A type YT direct shear/pullout tester for geosynthetics developed by Wenzhou Jigao Testing Instrument Co., Ltd. (Figure 1) was adopted in the following research. The test apparatus is mainly composed of four components: a vertical loading system, horizontal loading system, controlling—logging system, and the model box. The pullout model box is rectangular, which meets the design requirements of Test Methods of Geosynthetics (Ministry of Communications of the People’s Republic of China, 2006) for Highway Engineering (JTG E50-2006), with a 15-mm-thick bottom plate and two 10-mm-thick transparent toughened glass on both lateral sides, as well as a 5-mm-wide slit in the middle of the front end through which the geogrid could be connected with the clamp. The apparatus is highly automatic, comprehensively functional, and easy to operate. It is competent in direct shear or pullout tests for interfacial friction between various geosynthetic materials and different fillers. The main indexes of the tester are shown in Table 1.

The filler for this pullout test was prepared from Xiamen ISO standard sand. The standard sand was sieved in advance for preparation using an oscillating screener. The apertures of the sieves were 4.75, 2.36, 1.18, 0.6, 0.3, 0.15, and 0.075 mm, respectively. According to the specification requirements, the sieved sand samples with different particle sizes were first prepared into a well-graded filler S ₁, according to a specified proportion. Subsequently, the poorly graded fillers S ₂, S _3, and S ₄, which were mainly composed of coarse sand, medium sand, and fine sand, were prepared by the same means. The particle gradations and other basic properties of the four filler types are shown in Table 2.

The polyester warp-knitted geogrid (MacGrid WG 20) was used to investigate the drawing characteristics of toothed geogrid reinforcement. The physical and mechanical indexes of the geogrid are shown in Table 3. The rigid angle aluminum was adopted as the vertical tooth or denti-strip. The heights of the denti-strips were 5 mm and 10 mm, and the base of angle aluminum was drilled with a row of equidistant circular holes of 3 mm in diameter. The angle aluminum and the horizontal ribs of the geogrid were firmly fastened with tie wraps to form the qualified toothed geogrid. The length of the denti-strip was set equal to the width of the horizontal geogrid, and the toothed-geogrid sample was placed horizontally, as shown in Figure 2.

The pullout tests were carried out by combining the toothed geogrid with fillers S ₁, S ₂, S ₃, and S ₄. Considering different denti-strip heights and load levels, the effects of filler gradation on the pullout characteristics of toothed geogrids were observed. The prepared sand was constructed and compacted by four layers, controlling the weight and height of each layer according to the compactness of the test design. After the lower half of the model was constructed, the base surface was flattened and the toothed geogrid was then laid on it, ensuring that the denti-strip was located in the middle of the test chamber with symmetrical margins on both sides, and filling the upper half of the model upon the geogrid to complete the model installation. The normal loads were applied at three levels: 10, 20, and 30 kPa; the pullout velocity was set to be 0.5 mm/min. Additionally, the limit pullout displacement in this research was designated to be 20 mm. The test scheme is shown in Table 4, and the test procedure is given in Figure 3.

The coefficient of the reinforcing effect Q could be used to evaluate the enhancement of pullout resistance attributed by the denti-strip: Q = T max / T max ′ , (1) in which T _max and T′_max are the maximum pullout resistance of a toothed geogrid and a common geogrid within the observation range, respectively.

The interfacial quasi cohesion c* and the interfacial quasi friction angle φ* are often used to describe the friction characteristics of the reinforcement interface, as shown as follows: τ = c * + σ n ⁡ tan φ * , (2) τ = T max 2 L B , (3) where σ _n represents the normal stress (kPa) acting on the geogrid, τ is the interfacial friction strength (kPa), and L and B are the length and width (m) of the embedded geogrid, respectively. It should be noted that as a horizontal–vertical (H–V) reinforcement, τ hereby covers the resistance of the denti-strips.

The quasi-frictional coefficient f* is defined as the ratio of the interfacial shear strength to the normal stress σ _n : f * = τ σ n = c * σ n + tan ⁡ φ * . (4)

It considers the cohesion and, thus, presents comprehensive strength characterization rather than the pure friction effect.

The drawing coefficient K* is also often used to evaluate the interface characteristics between the geogrid and the filler (Ministry of Communications of the People’s Republic of China, 2012). According to the Technical Code for Application of Highway Geosynthetics (JGT/TD32-2012), it could be defined as follows: K * = tan ⁡ φ * tan ⁡ φ , (5)

in which φ is the friction angle of the filler.

Under the conditions of different denti-strip heights and particle gradations, the relationship between the pullout force (T) and pullout displacement (△L) of the toothed warp-knitted geogrid is given in Figure 4. At the initial stage of drawing, the pullout force at all levels of gradations rises linearly with the increase of pullout displacement and its inclination amplifies roughly with the increase of the coarse particle content. Subsequently, the pullout curve trend starts to differentiate. The curves under a higher external load (σ _n = 30 kPa) do not show obvious peaks and are characterized by strain hardening during the whole displacement observation extent. The increases of both normal load and denti-strip height will lead to enhancement of the pullout resistance.

The effect of filler gradation on the pullout force under different denti-strip heights presents a consistent pattern, i.e., the pullout resistance presents to be S ₄ < S ₃ < S ₁ < S ₂. It means that with the increase of coarse particle content, the pullout resistance magnifies gradually. The coarse particle content of filler S ₁ with a well-graded filler locates between S ₃ and S ₂; thus, the pullout force value is also located between the two. Since the coarse particle can cause a rougher interface for the reinforcement structure and provide a more significant interlocking effect which enhances the inter-particle occlusion and friction, the pullout force shows a certain degree of increase.

The coefficients of the reinforcing effect with different load levels and various denti-strip heights are given in Figure 5. It shows that all coefficients of the warp-knitted toothed geogrid are larger than 1.0, presenting a better reinforcing effect than the traditional geogrid sheet. On one hand, the coefficient of the reinforcing effect keeps growing as the denti-strip height increases. Although the denti-strip height would be constrained by many construction factors, a proper height of the tooth should be adopted and would be beneficial to the reinforcing effect. On the other hand, it is inferred that the coefficient of the reinforcing effect never reaches its largest when the highest normal stress σ _n = 30 kPa is applied on the test model. It denotes that the bearing resistance acting on the denti-strip might contribute a larger proportion in the total pullout resistance under lower load conditions. Thus, the bearing resistance acting on denti-strips is capable to play a more dominant role in certain cases such as low-backfill reinforcement. However, the coefficient of the reinforcing effect has no obvious correlation with the coarse particle content for all filler types. Furthermore, due to insufficient height and inadequate amount of the tooth in the warp-knitted toothed geogrid, the coefficients remain at a low level.

As shown in Figure 6, the interfacial shear strength parameters of reinforcement under each particle gradation can be obtained by linear fitting. In the figure, c ₅ * and φ ₅ * represent quasi cohesion and the quasi-frictional angle with the condition h = 5 mm. As indicated by the fitting equations, the values of the quasi cohesion c* and the quasi-frictional angle φ* show the same rule, that is, with the increase in the denti-strip height, both the parameters increase. Meanwhile, under the same denti-strip height, the values of the shear strength parameters increase with the increase in the coarse particle content. Under the condition of h = 10 mm, with the rise in the coarse particle content, the maximum increase of quasi cohesion (31.4%) from gradation S ₄ to gradation S ₂ was observed, while the quasi-frictional angle increases by about 15.8%. By contrast, under the S ₁ gradation level, with the rise of the denti-strip height, the maximum increase of the quasi-cohesion (12.8%) from h = 0 mm to h = 10 mm was observed, while the quasi-frictional angle increases by only 5.0%. The comparison shows that the effect of gradation and the denti-strip height on quasi-cohesion is much more significant than on the quasi-frictional angle.

The relation mesh of the quasi-frictional coefficient between the toothed warp-knitted geogrids and the sand is presented in Figure 7. Take the normal stress σ _n = 30 kPa as an example, the quasi-frictional coefficient f* of S ₁, S ₂, S ₃, and S ₄ are 1.29, 1.43, 1.18, and 1.12, when h = 0mm, respectively. When h = 10 mm, the quasi-frictional coefficient f* of gradation S ₁, S ₂, S ₃, and S ₄ are 1.43, 1.58, 1.31, and 1.22, respectively. The quasi-frictional coefficient increases with the rise in the coarse particle content under the same denti-strip height. With the same filler gradation, the quasi-frictional coefficient tends to decrease or increase with the growth of normal stress σ _n or tooth height h, respectively.

As shown in Figure 8, the drawing coefficient K* of the toothed geogrid shows an approximately linear increase with the growth of denti-strip height. Moreover, the pullout coefficients K* of the toothed geogrid are all larger than 1.8 under different test conditions, which shows that the toothed geogrid can greatly enhance the interfacial friction effect. With the same denti-strip height, the values of the drawing coefficient of the toothed geogrid are ranked as follows: S ₄ (fine sand with poor gradation) > S ₃ (medium sand with poor gradation) > S ₁ (standard sand with good gradation) > S ₂ (coarse sand with poor gradation). It could be seen that the order is just opposite to that of the quasi-shear strength parameters and quasi-frictional coefficient.

Generally, under the same test conditions, with the increase of the coarse particle content of the filler, the drawing coefficient gradually decreases, and the reasons can be attributed to as follows: 1) the reinforcement effect consists of two parts, namely, quasi cohesion and the quasi-frictional angle, while the drawing coefficient only reflects the enhancement of the latter; 2) for plain sand, the friction angle of coarse sand is usually much larger than that of fine sand; thus, the extent of further improvement after being reinforced is quite limited.