Appraisal of Anchor Arrangement and Size on Sand-Geogrid Interaction in Direct Shear

Soil–reinforcement interaction is a major factor in the analysis and design of reinforced earth structures. In current research the effects of attaching elements of different size and numbers as anchors on enhancement of interaction at soil–geogrid interface under direct shear conditions were studied. Poorly and well graded sands (SC & Sf), a high density polyethylene geogrid, anchors with three different size and numbers (layouts) and clamping length of 2 cm from shear surface were used. Samples were prepared dry at a relative density of 80% in a 30 × 30 cm direct shear box and subjected to normal pressures of 12.5, 25 and 50 kPa with the shear load applied at a rate of 1 mm/min. Results of the assessment show that anchored geogrids improve shear resistance at interface mainly due to mobilization of passive soil resistance that is significantly influenced by the magnitude of the normal pressure and the number and size of anchors. Interaction enhancements achieved varied between a minimum of 8% and a maximum of 42%.

required which are influenced by soil-reinforcement interactions. As enhancing interactions has attractions from economical and constructional perspective which are crucial factors in the design of reinforced earth structures many researchers including Chenggang (2004); Liu et al. (2014); Kargar and Mir Mohammad Hosseini (2016); Moghaddas Tafreshi et al. (2016); Tavakoli Mehrjardi and Amjadi Sardehaei (2017), Abdi and Safdari Seh Gonbad (2018) have investigated different methods of improving interactions. Soil-geogrid interactions are due to friction at interface, soil-soil shear strength within apertures and passive resistance in front of transverse ribs. Goodhue et al. (2001); Lopez (2002); Palmeira (2004); Abdi et al. (2009) and Xu et al. (2018) have shown that in direct shear mode the first two and under pull out condition the third mechanism is the main contributor to stability. As under direct shear mode, reinforcement tensile strength and passive resistance do not contribute significantly to stability of reinforced earth structures, this necessitates longer embedment lengths and larger volumes of soil, leading to greater costs. Thus, to alleviate this problem researchers have attempted to improve soil-geosynthetic interaction by different methods some of which are schematically shown in Table 1.
In current study the effects of soil grading, attached element size and arrangements together with normal pressures on interactions in direct shear mode have been assessed. Poorly and well graded sands, a geogrid type, three different element sizes, arrangements and normal pressures were adopted for the investigation.

Test Method
Large scale direct shear tests 30 × 30 × 17 cm were conducted on unreinforced, reinforced and anchored reinforced samples to determine the sands shear strength characteristics and the interaction parameters at soil-geogrid interface according to ASTM D3080-04 and ASTM D5321-08 respectively. Horizontal and vertical displacements were measured using LVDTs and a load cell was used for measuring shear forces together with an automatic data acquisition system.

Soils
Fine uniform and coarse non-uniform sands referred to from here after in the text and the figures by symbols S f and S c for abbreviation were used as soils (see Fig. 1). Sieve analysis was conducted on representative samples with the summary of the physical and the mechanical characteristics given in Table 2. According to the Unified Soil Classification System (USCS), the fine uniform sand (S f ) was classified as SP (poorly graded sand) and the coarse non-uniform sand (S c ) as SW (well graded sand). Figure 2 shows the high density polyethylene (HDPE) geogrid marketed as CE131 product of Meshiran representative of Netlon, UK used as the reinforcement together with its characteristics presented in Table 3. Figure 3 shows the plastic rosary beads of three different sizes and shape labelled A 1 , A 2 and A 3 and called anchors hereafter with their specifications presented in Table 4. These plastic elements were adopted because they already had holes through which wires for their attachment to geogrid surface could be passed.

Anchors
As under direct shear conditions, failure manifests as soil slippage on upper geogrid surface (Hsieh et al., 2011), consequently anchors were only attached to this surface. Based on the results of Abdi and Safdari (2018), for maximum interaction anchors were attached to the upper geogrid surface using wires 2 cm in length at approximately 45° to the horizontal soil-geogrid interface. Figure 4 shows the arrangements investigated according to which 6, 10 and 20 anchors were attached to each 30 × 30 cm geogrid specimen. The coding used for referring to the samples in the text, figures and the tables is S i A j L k , meaning "sand, S i ; anchor, A j and arrangement L k ".

Sample Preparation Method
To prepare samples, initially soils were placed on trays and completely dried under laboratory conditions. By adopting relative density of 80%, dry weight of soil required for filling the lower and the upper halves of the shear box in two layers were determined. After placing each soil layer in the lower and the upper halves of the shear box, they were compacted using a 10 × 10 cm metal plate and rubber pestle. The number of blows necessary for reaching the required relative density was predetermined by trial and error during preliminary experiments. The surfaces of the upper and the lower shear boxes at interface (shear surface) were covered with lubrication to avoid friction that would influence the test results.
In case of soil-geogrid and soil-anchored geogrid samples, to prevent displacement of the reinforcement during testing, it was initially positioned in the lower half of the shear box as shown in Fig. 5, and then sand was placed in two layers and each layer compacted. Subsequently the geogrid was laid back on the soil surface at the interface of the lower and the upper shear boxes before the upper half of the shear box was positioned, filled with soil layers and compacted. Subsequently the loading plate was placed on the soil surface in the upper shear box, LVDTs positioned, appropriate normal pressure applied and the shear load set at a rate of 1 mm/min. As direct shear failure usually manifests in the upper parts of reinforced earth structures where vertical pressures are relatively low (Palmeira, 2009), thus, normal pressures of 12.5, 25 and 50 kPa equivalent to 0.6-2.5 m height of soil were adopted. Tests were displacement controlled and loading continued up to a maximum horizontal displacement of 15 mm (i.e. 5% strain). At the end of each test, geogrids were exhumed and inspected for integrity and anchor detachment. In no case geogrid was damaged and in only 5 cases some anchors had been detached and needed replacement.

Enhanced Interaction Coefficient
Various researchers have defined the ratio of "reinforced soil shear strength to unreinforced soil shear strength" as interaction coefficient (C i ) which is used as a criterion for evaluating soil-geosynthetic interaction (Tatlisoz et al., 1998;Abdi & Arjomand, 2011). In current research, "enhanced coefficient of interaction (C ie ) has been introduced which is defined as the ratio of "soil-anchored geogrid shear strength to soil-geogrid shear strength". This coefficient denotes improvement in soil-geogrid interaction due to the attachment of various numbers of anchors as follows: where: δ e = soil-anchored geogrid interaction angle (°); δ a = soil-geogrid interaction angle (°); C e = apparent adhesion at soil-anchored geogrid interface (kN/ m 2 ); C a = apparent adhesion at soil-geogrid interface (kN/m 2 ); σ n = applied normal pressure (kN/m 2 ). As cohesion less soils were investigated, the above equation simplifies to: (1) C ie = C e + σ n tan e C a + σ n tan a

Results and Discussion
Large scale direct shear tests were conducted on unreinforced, reinforced and anchor-reinforced samples, results of which are presented below as variations of shear stress-shear displacement and shear strength envelopes.

Unreinforced Samples
Result of tests carried out on samples of unreinforced sands S c and S f are shown respectively in Figs. 6 and 7, and were used as a criterion for evaluating interaction enhancements. It is observed for both sets of samples that shear stresses initially increase with shear displacement (i.e. hardening behaviour) and after reaching failure state (i.e. maximum value) reduce with further displacement displaying softening behaviour which is typical of granular materials. The rate of change of shear stress with shear displacement for all samples are relatively the same up to 5 mm displacement and the main differences are in the maximum shear stress values and the corresponding shear displacements. Samples of sand S c with non-uniform and coarser-grains consistently displayed higher shear stress values compared to S f samples subjected to the same normal pressures. Sand S f samples reach their maximum and ultimate states respectively at shear Internal angles of friction of S c and S f sands were determined respectively as 44.8 and 33.8 degrees, with negligible apparent cohesions (Fig. 8).

Reinforced Samples (S c G & S f G)
Figures 9, 10, 11 show variations of shear stress-shear displacement and failure envelopes obtained for reinforced S c G and S f G samples respectively. Considering Figs. 9 and 10, it can be seen that by reinforcing sands S c and S f samples, trend of variations of shear stresses with shear displacement are similar to behaviour observed for unreinforced samples with slightly smaller maximum shear stresses, particularly at σ n = 50 kPa. Samples show initial hardening and subsequent softening behaviour after failure supporting observations of Liu et al. (2009) and Mossallahnezhad et al. Mosallanezhad et al. (2016). The reduction in shear strengths is a clear indication of poor soil-geogrid interaction under direct shear mode. These changes are attributed to soil particle slippage on upper geogrid surfaces without its tensile strength and soil passive resistance being mobilised as well as the reduction of soil-soil interfaces also reported by Cazzuffi et al. (1993); Alfaro et al. (1995), Abu-Farsakh et al. (2007); Liu et al. (2009) and Kim and Ha (2014). Failure envelopes for the reinforced as well as the unreinforced samples for comparative purposes are presented together on Fig. 11. Interaction angles measured for the reinforced S c and S f samples were respectively 39.2 and 28.3° showing 13 and 16.3% reduction compared to the unreinforced samples. Smaller reduction in angle of interaction of reinforced S c compared to S f samples is attributed to its nonuniform, larger and angular particles which interacted more effectively with each other and the geogrid members. Apparent adhesions for reinforced samples were negligible and disregarded in accord with the results reported by Goodhue et al. (2001) and Athanasopoulos (2002). They attributed the small apparent adhesions to soil suction at particle contacts as well as friction intrinsically built in the direct shear equipment.

Anchor-Reinforced Samples (S c A j L k & S f A j L k )
To improve soil-geogrid interaction at interface in direct shear condition and compensate for the loss of shear strengths as a result of reinforcing, elements with three different sizes and arrangements (i.e. layouts) having an anchorage length of 2 cm were attached to the upper geogrid surface and tested. Because of the variety of samples only the results for S c A 1 L 1 and S f A 1 L 1 samples that resulted in the greatest improvements are shown on Figs. 12, 13, 14, 15.
For comparative purposes results of reinforced samples are also included.
Considering Figs. 12 and 14 it can be observed that for both sets of samples subjected to low normal pressure of 12.5 kPa, attachment of anchors to geogrids has not resulted in noticeable improvement in shear stresses at interface in comparison to equivalent reinforced samples. By increasing normal pressures to 25 and particularly 50 kPa, attachment of elements to the geogrids has resulted in significant improvement in maximum shear stresses and reduction of corresponding shear displacements. In both sets of samples subjected to σ n = 50 kPa the rate of change in shear stresses with shear displacement have significantly increased compared with samples subjected to  14 Shear stress-shear displacement for S f G and S f A 1 L 1 samples σ n = 12.5 kPa, clearly indicating that soil-anchored geogrid interaction are dependent on the magnitude of the normal pressures. The improvements are credited to greater particle confinement, increased soil-geogrid, soil-anchor and soil-soil interactions as well as passive resistance mobilization in front of the anchors.
At the early stages of loading (< 2 mm shear displacement), shear stresses for reinforced and anchor reinforced samples particularly those subjected to σ n = 12.5 and 25 kPa are very similar and the curves almost overlap. As loading continues the rate of change of shear stresses with shear displacements gradually decrease. After reaching maximum values (i.e. failure state) shear stresses gently decrease and samples reach ultimate states at shear displacements of 5-10 mm depending on the normal pressure. Generally the higher the normal pressure, the greater the maximum shear displacement at which ultimate state is reached. Anchor reinforced samples reach failure states at smaller shear displacements than equivalent reinforced samples and mobilise shear strength at a faster rate particularly those samples subjected to σ n = 50 kPa. As an example, by anchoring S c G sample (i.e. S c A 1 L 1 ) subjected to σ n = 50 kPa maximum shear stresses increased from 44.51 to 63.23 kPa, which is an improvement of 42 percent (Fig. 12). Enhancements in soil-geogrid interaction by the addition of metallic angles to the reinforcement in pull out tests have also been reported by Sadat Taghavi and  and Mosallanezhad et al. (2017).

Failure Envelopes
Summary of all test results as maximum shear strengths (τ max ), interaction coefficients (C i ), enhanced interaction coefficients (C ie ), internal friction angles (Φ), interaction angles at soil-geogrid interface (δ a ) together with soil-anchored geogrid interaction angle (δ e ) are presented in Table 5.

Unreinforced and Reinforced Samples
Considering Table 5, it can be seen that angles of internal friction determined for the unreinforced S c and S f samples were 44.8° and 33.8° which were reduced to 33.8° and 28.3° respectively as a result of reinforcing with geogrid (i.e. S c G & S f G) giving interaction coefficients of 0.88 and 0.87 respectively. The reduction in interactions is attributed to slippage of soil particles on upper surfaces of geogrid elements, inadequate passive soil resistance and reduction of overall soil-soil particle interfaces.

Anchored Geogrid Samples
Summary of the results presented in Table 5, clearly show that reinforcing sands with geogrid under direct shear mode reduces interactions and the attachment of elements to the reinforcement surfaces significantly improves interaction at interface. Attachment of different sizes and arrangement of anchors to geogrid surfaces in both sets of samples not only compensated the reduction in shear strengths, in fact in most cases soil-geogrid interaction angles were enhanced to values greater than sands friction angles. Highest C ie values of 1.42 and 1.35 showing respectively 42 and 35% improvement in interactions were achieved for S c A 1 L 1 and S f A 1 L 1 samples. For the same arrangement of anchors (L 1 , L 2 or L 3 ), the larger elements with greater base and surface areas (i.e. A 1 > A 2 > A 3 ) consistently produced the highest enhancements which signifies the importance of passive resistance mobilization on interactions. It is observed that for the coarser non-uniform sand (S c ) the attachment of even 6 of the smallest elements (i.e. S c A 3 L 3 ) has almost compensated the reduction in soil shear strength due to reinforcing under direct shear Fig. 15 Failure envelopes for S f , S f G and S f A 1 L 1 samples mode whereas for the finer uniform sand (S f ) at least 10 of the same elements are required (i.e. S f A 3 L 2 ). This behaviour clearly emphasises the important rule of particle grain size and distribution on soil-geogrid interaction.
According to the results enhanced coefficients of interactions (C ie ) obtained for both S c and S f sands subjected to normal pressures of 12.5 and 25 kPa do not differ significantly and are almost the same as interaction coefficients (C i ). This means that attachment of elements to geogrid reinforced samples subjected to low normal pressures do not lead to significant improvements in soil-geogrid interactions. By increasing the normal pressure to 50 kPa, enhanced coefficients of interaction show significant changes and vary between a minimum of 1.16 for S C A 3 L 3 and a maximum of 1.42 for S c A 1 L 1 respectively and 1.08 and 1.35 for S f A 3 L 3 and S f A 1 L 1 samples. These represent enhancements of 16-42% for S c with D 50 = 1 mm and 8-35% for S f samples with D 50 = 0.22 mm. Thus it is concluded that normal pressure is a more influential factor on enhancing soil-geogrid interaction than particle size and distribution. Figures 16 and 17 clearly show that for a particular sand (i.e. S c or S f ) and element layout (i.e. L 1 , L 2 & L 3 ) the highest improvements in interaction have consistently been achieved for the samples with the largest anchors (i.e. A 1 ). Results also show that attachment of even 6 anchors to the geogrids (i.e. L 3 = 6) improved anchored interaction friction angles (i.e. δ e ) to those of soil internal friction angles. These observations indicate that the larger the size and the number of attached elements to the geogrid surface, the greater the mobilization of passive soil resistance in front of these elements and thus the higher and the more effective the soilgeogrid interaction.

Conclusions
Reinforcing soils under direct shear mode result in the reduction in shear strength due to poor soil-geogrid interaction. In present research, improvement of soil-geogrid interaction by mobilizing passive soil resistance using elements as anchors was investigated. Effects of influential factors such as anchor size and layout (numbers), soil particle size and distribution together with normal pressure on interaction were assessed and the following conclusions drawn: • Reinforcing coarse grained non-uniform (S c ) and fine grained uniform (S f ) sands with geogrid under direct shear mode results in the reduction of soilgeogrid interaction characteristics at interface and thus overall shear strength. • Attachment of elements as anchors to geogrid surface significantly enhances shear strength characteristics at soil-geogrid interface in case of both coarse grained non-uniform (S c ) and fine grained uniform (S f ) sands shown by higher interaction angles (δ e ) achieved compared with soil internal friction angles (φ). • Interaction enhancements are influenced by anchor size and numbers (i.e. layouts) as well as magnitude of normal pressure. Anchors with the larger cross section and greater numbers result in greater soil-anchor interaction and mobilization of passive soil resistance. • Normal pressure due to the confinement effects it provides is shown to be a more influential factor on enhancing soil-geogrid interaction than soil particle size and gradation. • Attaching anchors to geogrids resulted in improving interaction angles in S c samples by a minimum and a maximum of 4.5 and 10.9 degrees (i.e. 10 to 24.3%) and in S f samples by 2.1 and 9.7 degrees (i.e. 6.2 to 28.8%) in comparison to unanchored reinforced samples. Minimum enhancements correspond to samples with the lowest number and the smallest anchors (i.e. L 3 & A 3 ) and the maximum to samples with the largest number and the largest anchors (i.e. L 1 & A 1 ).
Funding The article has had no funding.
Data Availability Enquiries about data availability should be directed to the authors.

Conflict of Interest
The authors have not disclosed any competing interests.