Ecological Retaining Wall for High-Steep Slopes: A Case Study in the Ji-Lai Expressway, Eastern China

Research on the retaining structures for high-steep slopes is extremely significant because of its real-world applications and farreaching implications. A flexible geocell-reinforced ecological retaining wall as a high-steep slope protection scheme was developed and applied to the slope protection project of the Ji-Lai Expressway by analyzing the reinforcement mechanism of the geocell used. ,e lateral displacement and Earth pressure distribution on the flexible ecological retaining wall applied to the highsteep slope were studied using finite element numerical simulations and verified using field experiments. Results reveal that the wall maximum horizontal displacement is 2/3 H away from the wall toe because of the replacement of the upper part of soil. ,ere is an obvious bucking effect on the active Earth pressure around the stiffened site, and the flexible deformation of the retaining wall helped effectively release some of the Earth pressure. Consequently, the measured value is lower than the theoretical value. ,rough this case study, it is demonstrated that the flexible ecological retaining wall as a slope protection technology can be successfully applied to steep slopes with a height of more than 15m. Moreover, it brings significant advantages for protecting the ecological environment and improving the highway landscape.


Introduction
High-grade highway slopes are built in the process of filling embankments or during excavations to improve highway networks. To stabilize such slopes, surface protection measures are implemented, such as grout coating, concrete coating, and mortar rubble coating, and common measures, such as flexible retaining walls and rigid retaining walls, are applied to retain a reinforced structure [1,2].
Many types of rigid retaining walls have been employed to retain stabilization of slope, including cantilever retaining walls and gravity retaining walls. To promote the application of dry-stone walling to current civil engineering practices, Colas et al. [3] conducted a failure test on a full-scale model, validated and calibrated a yield design model, and evaluated the stability of a dry-stone retaining wall. However, from the perspective of building materials and methods, the technical requirements for such structures are less, and practical applications are bound to have some limitations. Choi et al. [4] and Saim and Kasa [5], respectively, conducted experimental tests and numerical simulations on a T-shaped cantilever retaining wall and studied its displacement mode and stress state. Limited by its height, the T-shaped cantilever retaining wall could not contribute to the protection of the high-steep slopes. By conducting a scale model test, Zhu et al. [6] studied the factors influencing the deformation mode of a gravity retaining wall and the failure mode of footings made of different materials. Liu et al. [7] employed the pseudostatic method and limit equilibrium method to study the effect of embedded depth on the critical acceleration of a soil-wall system and quantified the seismic rotational stability of a gravity retaining wall. ey showed that the embedded depth of the footing significantly influences the stability of the wall: they emphasized on using a deep footing for gravity retaining walls intended for high-steep slope protection. Clearly, a rigid retaining wall is unsuitable for protecting high-steep slopes because of certain limitations like the self-weight effect, deep footing, high structural requirements, and huge construction cost.
Methods for slope protection are usually detrimental to the original vegetation and restoration efforts, resulting in a significant amount of secondary bare land, and the subsequent rainfall and runoff energy can lead to soil erosion and, in some cases, slope collapse [8], which will further degrade the ecosystem. Vegetation significantly affects the hydrological and mechanical properties of hill slopes, which are conducive to the long-term stability of the slope ecosystem [9]. Sanchez-Castillo et al. [10] concluded that in addition to studying the technical factors of plants as structural components, the ecological and morphological characteristics of the different species and the adaptation of plants to the disturbance should be considered. Based on the limit equilibrium principle, Bischetti et al. [11] improved the existing mechanical model that provided the coefficient of safety for a specific slope, quantifying the effect of brush layering on slope stability. Yan et al. [12] defined the concept of original ecological tridimensional slope vegetation and studied the slope fixation mechanism of vegetation by performing a theoretical and observational analysis and a comparison of tridimensional slope vegetation in the wild.
From the perspective of protecting the ecological environment, enhancing the highway landscape, and maintaining an ecological balance, the application of soil bioengineering technologies should be encouraged. Cao et al. [13] adopted air bricks for slope protection, where the holes in the air bricks were filled with vegetation, providing a good example for increasing vegetation cover through ecological engineering. However, the transportation of air bricks is inconvenient because of their overall weak structure, and there remain many limitations for steep slope protection. Compared with air bricks, geocells not only provide space for planting vegetation, thereby improving the amenity values of the surrounding environment, but also conform to the terrain owing to their flexibility, convenient transportation, and construction scheme. Barring the effect of plant solidification, the confinement of a three-dimensional (3D) geocell itself can improve the bearing capacity of soil and prevent soil erosion [14,15].
Flexible geocell-reinforced retaining walls have drawn considerable research attention in recent years. Currently, the stability of such walls can be studied using methods such as field monitoring, centrifugal model testing [16], and theoretical simulation and calculation [17]. Pokharel et al. [18] experimentally studied the factors influencing the stiffness and bearing capacity of single geocell-reinforced retaining walls. Chen and Chiu [19] carried out a model test to analyze the reinforcement action and the failure mechanism under an overload. Leshchinsky et al. [20] analyzed the slip surfaces of geocell-reinforced retaining walls through shaking table tests and recommended the equivalent pseudostatic coefficients in the earthquake-resistant design of reinforced retaining walls. By conducting centrifuge model tests, Song et al. [21] found that the failure mechanisms of geocell-reinforced retaining walls were similar to those of slopes.
Clearly, many studies have been conducted on the stability of geocell-reinforced retaining walls under various simulated loads. However, there are some gaps between practical application and theoretical analysis with respect to geocell-reinforced retaining walls under different geological conditions; the study of the effects of lateral Earth pressure and displacement distribution might as well be regarded as an important research topic. In this study, based on a structural analysis and finite element numerical simulation of a geocell-reinforced retaining wall, a high-steep slope protection scheme is applied to a highway slope of the Ji-Lai Expressway. e simulation results were verified using field test monitoring data, and the lateral Earth pressure and displacement distribution were studied. ese results can serve as a basis for further research on geocell-reinforced retaining walls as a highway slope protection scheme.

Project Background.
e section of the Ji-Lai Expressway that was evaluated is located on the second-level terrace of a river valley. e slope of the terrace is nearly vertical, and the height of the terrace is in the range of 15.47-16.94 m (Figure 1(a)). e foundation soil is Q3 collapsible loess. It is common knowledge that loess with some typical internal structure is found to be more stable than common soil; however, a field survey has pointed to the presence of cracks in this soil layer, causing some damage to the subgrade. Penetrating longitudinal cracks over 3 m wide can be observed on the roadbed surface, as shown in Figure 1(b). e edge of the roadbed has collapsed, and some transverse cracks have intruded 0.92 m along the roadbed. Any highway engineering project performed near this geological structure could seriously affect the stability of the embankment.

Retaining Wall Scheme.
To eliminate hidden dangers and ensure highway engineering quality and driving safety, the necessary engineering measures for slope protection need to be evaluated. e commonly used rigid gravity retaining wall scheme was voted down for the following reasons at the early stage of this project: (1) the JTG D30-2004 Specifications for Design of Highway Subgrades [22] state that gravity masonry retaining walls should not exceed a height of 12 m; considering the embedded depth of the footing and main body structure, the wall height of a gravity masonry retaining wall would exceed 18 m, which is far beyond the specified value; (2) the heavy weight of masonry retaining walls has more requirements in terms of construction technology; hence, there will be no space for ecological greening; (3) in engineering practice, the construction cost is a key factor that should be considered. e estimated construction cost based on earthwork processing, material cost, and foundation treatment is approximately 763,000 yuan.
An ecological geocell-reinforced retaining wall scheme was proposed owing to the following reasons: (1) a geocellreinforced retaining wall is lighter than a masonry structure, with only minor foundation treatment required, and the structural damage to the retaining wall due to the uneven settlement of the footing can be avoided; (2) the vegetation cover around the geocell on the retaining wall surface can help resist soil erosion and slope instability, which plays a role in afforestation; (3) the construction cost is relatively low, with the estimated construction cost based on earthwork processing, material cost, and foundation treatment being approximately 542,000 yuan.
Given the many advantages of flexible geocell-reinforced retaining walls, the construction scheme is determined as follows. e main structure is divided into two parts: a 3 m high lower part resting on a 4.8 m wide mortar rubble footing and a 12 m high upper part resting on a 4 m wide geocell-reinforced retaining wall with a slope gradient of 1 : 0.5. Moreover, above the retaining wall is a 3 m high filling slope with a gradient of 1 : 1.5. Figure 2 shows the retaining wall structure in greater detail.

Reinforcement Mechanism.
Studying the reinforcement mechanism of geocells is conducive for practical engineering applications. Under the action of the upper load, the geocell wall provides a good lateral restraint, which is controlled by the tensile strength of the geocell material. e volume of the geocell is assumed to remain unchanged, and according to the rubber membrane theory proposed by Henkel and Gilbert [23], the additional confining pressure σ * due to the membrane stresses can be expressed as where M is geocell material modulus, [MPa], d 0 is initial diameter of an individual geocell pocket, [m], and ε 0 is geocell allowable axial strain. Under this pressure, anchorage (or interfacial friction, or both) is imposed by the geocell walls ( Figure 3). us, the geocell comprehensively behaves as a semirigid composite slab with the surrounding soil. e reinforced layer forms a concave surface under the action of the upper load, and the pressure P on the concave surface is greater than the pressure P b beneath the concave surface, thus generating a net effect. e equivalent modulus E * of a semirigid composite slab is improved by the geocell, according to the principle of deformation equivalence: where E is soil deformation modulus (MPa) and P s is upward stress provided by the geocell wall (kPa). rough the net effect and owing to the slab configuration, the upper load is uniformly redistributed to a wider area, with a lower intensity to improve the load-bearing capacity of the soil, and to a certain extent, the differential settlement under the action of the upper load is effectively resisted [24,25]. e apparent cohesion of sand, even of dry sand, is close to zero. A geocell induces a significant apparent cohesion on the infilled soil [26], and the walls cut the potential slip surface (Figure 4), which develops in an unreinforced condition and forces it to go deeper into the soil, thereby increasing the stability and bearing capacity of the soil [16,27]. To summarize, the mechanism of geocell reinforcement can be interpreted as "confinement," "membrane action," and "stress distribution" [25,28]. e above qualitative analysis demonstrates that the geocell helps improve the lateral constraint on the reinforced soil, and the deformation of the soil is decreased. In addition, owing to the lifting force of the geocell, the deformation of the lower part of the soil is reduced. In fact, the overall role of the geocell in the subgrade is much more complex. To apply and verify the above reinforcement theory in engineering practice, several indoor and outdoor tests should be conducted to accumulate the measurement data.

Physical Parameters.
e main physical and mechanical properties of the loess in the section of the Ji-Lai Expressway taken as the test case should be determined, which is conducive to the theoretical calculation, numerical simulation, and practical engineering application ( Figure 5). In accordance with the JTG E40-2007 Test Methods of Soils for Highway Engineering [29], the loess was subjected to a Advances in Civil Engineering number of tests, namely, plastic limit, liquid limit, compaction, compression, shear, and particle analysis tests. A repeatability test helps control the error, and the test data are recorded. Test results are listed in Table 1. e geocell specifications are listed in Table 2.
e following assumptions are made before conducting the theoretical calculation and analysis:  (c) e tension design value of each layer is determined by the pullout resistance T i (kN) of the geocell stiffener, where i is the number of layers. e pullout resistance can be calculated as follows: where S i is ultimate pullout capacity of the ith layer (kN); c is bulk density of the backfill (kN·m − 3 ); and [K f ] is dimensionless safety coefficient for pullout stability. (d) e filling slope gravity and vehicle load acting on the retaining wall are converted to the equivalent soil layer thickness h (m): where h 1 is filling slope height above the retaining wall [m]; B v and L 0 are load arrangement length and width, respectively (m) (the width of subgrade); and G v is wheel load or crawler load acting in the area of B y L 0 (kN). e Earth pressure Ea (kN) acting at the back of the retaining wall and the height of the net force working point H E a (m) are calculated as follows: where K a is the dimensionless coefficient of the active Rankin pressure, K a � cos 2 (φ − α)/cos 2 α cos (δ + α) [1 + ����������������������������� � sin(φ + δ)sin φ/cos(δ + α)cos α ] 2 ; δ is friction angle at the back of the retaining wall (°); α is retaining wall inclination (°); φ is geocell soil internal friction angle (°); and H is retaining wall height (m).

Feasibility Analysis.
e stability of the retaining wall should be analyzed considering the resistance provided by the retaining wall itself and the resistance of the geocell stiffener. Figure 6 shows the theoretical calculation schematic applied to the flexible retaining wall.

Advances in Civil Engineering
where μ is soil dimensionless Poisson's ratio; N is axial force on the footing (kN); E y , E x are earth pressure vertical and horizontal components, respectively (kN); and G is retaining wall gravity (kN). (2) Analysis of antioverturning stability: where B is geocell-reinforced retaining wall width (m); e' is eccentricity at the retaining wall bottom (m); e is eccentricity at the footing bottom (m); B f and b f are footing width and height, respectively (m); G f is footing gravity (kN); σ max and σ min are maximum and minimum compressive stresses of the footing bottom, respectively (kPa); σ is revised allowable bearing capacity of the foundation soil (kPa); and k is dimensionless allowable bearing capacity improvement coefficient of the foundation soil. e overall sliding stability of the retaining wall should be analyzed with the assumption that the geocell stiffener length should not exceed the displacement length of the sliding surface, where it may occur. e Fellenius method of slices can be used for the calculation: Figure 6: eoretical calculation of the flexible retaining wall. 6 Advances in Civil Engineering where c i , L i are cohesion and arc length on the ith block sliding surface, respectively (kPa) and (m); W i is dead weight and its applied load on the i block (kN); φ i is i block soil internal friction angle (°); and α i is angle between the normal and the vertical line of a sliding arc of the i block (°). After the calculations are made using the above formula, the flexible geocell-reinforced retaining wall scheme meets all the theoretical requirements.

Numerical Parameters.
Backfills and retaining walls are widely regarded as ideal elastoplastic bodies and ideal linear elastomers subjected to the Mohr-Coulomb failure criterion; therefore, a constitutive model based on the Mohr-Coulomb criterion is adopted for the analysis, which can be expressed as follows: where J 2 is the second invariant of the deviatoric stress tensor; θ α is lode angle (− 30°≤ θ α ≤ 30°) (°), which is the angle between the first principal stress and the deviatoric stress component; and I 1 is the first invariant of the stress tensor. e geocell-reinforced soil triaxial compression test conducted by Chen et al. [30] indicated that reinforced soil exhibits an obvious cohesion; however, the friction angle of the reinforced soil is almost the same as that of the unreinforced soil. erefore, the internal friction angles of the unreinforced soil and geocell-reinforced soil are assumed the same. e cohesion of geocell-reinforced soils can be estimated using an empirical equation developed by Bathurst and Karpurapu [31] and Rajagopal et al. [26] as follows: where K p is dimensionless coefficient of passive Rankin pressure; K p � tan 2 ((π/4) + (π/2)); and σ * is additional confining pressure due to the membrane stresses in equation (1). Furthermore, Madhavi Latha et al. [32] and Madhavi Latha and Rajagopal [33] verified the correctness of this equation. From the perspective of system engineering, the flexible ecological retaining wall, geocell stiffener, rubble footing, and backfill contribute to the maintenance of subgrade stability. ey interact with each other, redistribute, and transfer stress dynamically as the load changes. A numerical simulation through ABAQUS (Dassault Systèmes 2017) can help strengthen and improve its stability design. To simplify the model and improve the computational efficiency, a two-dimensional (2D) composite model with equivalent strength and stiffness is adopted, which is accepted in the engineering community [30,34]. e calculation area is divided into several sections: flexible retaining wall, geocell stiffener, rubble footing, backfill, and foundation. Based on the physical parameters measured through laboratory testing and the feasible range of parameter values in the JTG D30-2004 Specifications of Design of Highway Subgrades [22], the numerical parameters can be determined using equation (15), which are summarized in Table 3. Geocell was modeled as a 2-node linear truss element [35,36]. A surface-to-surface contact is employed to define the interfaces. Chen et al. [37] indicated that the interface friction angle between two geocell-reinforced sand layers is equal to the friction angle of sand because the thickness of the geocell wall was small enough. erefore, the friction angle is specified as 25°. A tie constraint is set at the interfaces between the footing and foundation and between the wall and geocell [38].

Simulation Model Description and Configuration.
In the process of nonlinear calculation, the final and most important step before applying the loads is to determine the soil initial stress state, which affects not only the initial loading calculation but also all the subsequent loading steps.
is determines the success of the numerical simulation, as the stress in the subsequent loading steps is also accumulated on the basis of the previous initial stress. Considering the initial geostress balance [39,40], first, the gravity load is applied to the finite element model, and the stress field calculated under the gravity load is then defined as the initial stress field and applied to the original model along with the gravity load.
us, both the equilibrium condition and the initial stress field of the yield criterion are satisfied, and it can be ensured that the initial displacement of each node is approximately zero.
Finally, a finite element simulation is conducted after the corresponding loads (the 3 m high filling slope above the Advances in Civil Engineering 7 retaining wall with a slope gradient of 1 : 1.5 is regarded as the dead load) are applied to each section.

Field Test Arrangement.
e actual distribution of soil pressure and the change in the retaining wall horizontal displacement are determined to verify the numerical calculation results and obtain the working characteristics of the geocell-reinforced retaining wall in practical engineering. Based on the layout of the test elements shown in Figure 2, the pressure cells were buried at the middle and back of the flexible retaining wall, and displacement-monitoring points were laid on the surface of the retaining wall for long-term monitoring and data collection. e field monitoring device calibration and placement process is shown in Figure 8.
As shown in Figure 2, in the cross section of the flexible ecological retaining wall, ○ represents the buried position of the pressure cells and ▽ represents the displacement observation points. Pressure cells numbers 1-5 were embedded laterally at the middle of the wall (separated 2 m from the retaining wall surface); pressure cells numbers 6-10 were embedded laterally at the back of the wall (separated 4 m from the retaining wall surface). e displacement observation points were set at the surface of the retaining wall.   Advances in Civil Engineering e soil pressure test results might be affected by the physical environment, construction disturbance, and quality of the test elements, resulting in a greater dispersion of the soil pressure. In this study, the value with greater dispersion was removed during the analysis of the soil pressure test results. Figure 9(a) shows the lateral displacement distribution obtained from the numerical model. e maximum horizontal displacement of the retaining wall is approximately 1/3 H away from the wall   Advances in Civil Engineering toe. In the Rankine Earth pressure theory, the positive active Earth pressure of the cohesive soil is due to the dead and external loads, and the negative active Earth pressure is due to the cohesive force. According to equation (6), the net force of the active Earth pressure acts (H(H + 3h)/3(H + 2h)) away from the wall toe. e horizontal component of the displacement vector of the retaining wall, described in Figure 9(b), can intuitively reflect the effect of the net force, consistent with Rankine Earth pressure theory. Figure 10 shows the variation in measured displacement data from the field test with consolidation time: the horizontal displacement is maximum 2/3 H away from the wall toe. is is because the backfill soil in the top 3 m range of the retaining wall was replaced by lime soil to prevent and control the existing longitudinal cracks in the actual project. Under horizontal active Earth pressure, the disturbed soil layer generates a larger horizontal deformation. Because the backfill soil behind the lower part of the retaining wall is undisturbed soil, the horizontal deformation is limited, and the deformation is the highest at the upper part, that is, approximately 2/3 H away from the wall toe, while the simulated displacement is maximum approximately 1/3 H away from the wall toe. is indicates that the complexity of the actual construction environment makes it difficult to evaluate the numerical parameters.

Lateral Displacement Analysis.
Although the flexible retaining wall could conform to larger deformation, the maximum horizontal displacement measured does not exceed 42 mm because the entire soil structure inside the retaining wall is not damaged. e lateral deformation due to gravity is negligible, and only small lateral deformation occurs under the action of the roadbed structure and traffic loads. Moreover, as shown in Figure 10, the deformation is stable 180 days after completion. Compared with the deformation 210 days after completion, there is little difference between them. Figure 11 shows four selected cross section stress distribution profiles. As we move farther away from the retaining wall surface, the lateral Earth pressure along the height of the retaining wall presents more pronounced fluctuations. At the same horizontal position, the simulated value at the middle of the wall is greater than the measured value at the back of the wall (closed symbols, no line in Figure 11). is is because the geocell stiffener provides an opposite constraint, yielding a bucking effect on the active Earth pressure around the stiffened site. e closer to the stiffened site, the more obvious the restraint effect. erefore, the measured value at the back of the wall is less than that at the middle of the wall.

Lateral Soil Pressure Analysis.
e measured values at the middle and upper parts are in good agreement with the simulated value ( Figure 11); however, there is a significant difference between the measured and simulated values at the lower part of the retaining wall. All the backfill behind the wall is disturbed soil in the numerical simulation. In practice, only the upper part behind the wall is disturbed soil with unstable mechanical properties, while the lower part is undisturbed soil with undamaged natural structure and good mechanical properties. Consequently, the lateral Earth pressure is not as high as the simulated value, which is also evident from the measured value on different days after completion, as shown in Figure 12.
e theoretical value of the lateral Earth pressure can be obtained using (5), and as a comparison, the measured Earth pressure is also lower than the theoretical value (see Figure 12). Because of the flexible deformation of the wall, some amount of the Earth pressure is fully released; hence, the actual lateral pressure coefficient of the backfill is lower than the theoretical value. As shown in Figure 11, the measured Earth pressure is maximum approximately 1/3 H away from the wall toe, which is the height of the net force working point H E a in the Rankine Earth pressure theory. Compared with the lateral Earth pressure 10 days after completion, the lateral Earth pressure settles more than six months after completion.
is shows that under the action of load, soil consolidation is complete, and the mechanical properties of each part of the soil play a uniform role.   is case study demonstrated the effectiveness of flexible ecological retaining walls as a slope protection technology for high-steep slopes with a height of more than 15 m; moreover, it brings significant advantages in protecting the ecological environment and improving highway landscape.

Conclusions
In this study, a flexible geocell-reinforced ecological retaining wall as a high-steep slope protection scheme was developed and applied to a portion of the Ji-Lai Expressway by analyzing the reinforcement mechanism of the geocell used.
e lateral displacement and Earth pressure distribution on the flexible ecological retaining wall applied to the high-steep slope were studied using theoretical calculations  and a finite element numerical simulation and verified using field experiments. After a detailed analysis of the results, our main findings are outlined as follows: (1) e simulated horizontal displacement is maximum approximately 1/3 H away from the wall toe, while the measured displacement is maximum 2/3 H away from the wall toe. e deformation is maximum in the middle and upper parts, and the maximum measured horizontal displacement does not exceed 42 mm.
(2) With the increase in the distance from the retaining wall surface, the lateral Earth pressure along the height of the retaining wall presents more fluctuations. e measured value at the back of the wall is lower than that at the middle of the wall. (3) e measured value in the middle and upper parts is in good agreement with the simulated value; however, there are big differences between the measured and simulated values at the lower part of the retaining wall. is indicates that the complexity of the actual construction environment makes it difficult to evaluate the numerical parameters. (4) e lateral Earth pressure is maximum approximately 1/3 H away from the wall toe. e measured lateral Earth pressure is less than the theoretical value of the active Rankine pressure. e geocell-reinforced slope protection technology is an effective method to retain slope stability. (5) Soil consolidation is stable, and the mechanical properties of each part of the soil tend to be uniform six months after construction. e proposed approach brings significant advantages in protecting the ecological environment and improving highway landscape. (6) Relatively speaking, the finite element simulation result was not the same as actual value due to the complexity of actual construction progress. erefore, adequate and effective field survey and intensive structure division should be suggested to further consider in similar future studies.

Data Availability
e data used to support the findings of this study are available from the corresponding author upon request.

Conflicts of Interest
e authors declare that there are no conflicts of interest regarding the publication of this paper.