Loess covers an area of about 6.3×105 km2 in Northwest China, accounting for 6.6% of the land area. Induced by precipitation, evaporation, uneven settlement, or external load, tensile cracks emerged in loess slopes or foundations when the tensile strength of loess was exceeded, and further affects the mechanical and hydraulic properties [34, 38, 40, 44]. Due to the occurrence of tensile cracks, massive geohazards occurred every year in the Loess Plateau of China, such as landslides and ground fissures [24, 28, 33, 45, 46]. The above geohazards are closely related to the lower tensile strength of loess, which necessitates an effective method to improve the strength and resist cracking of loess.
The addition of discrete and randomly distributed fibers can well improve the tensile strength and ductility of the soil [26, 36]. The reinforcement effect of fibers for the mechanical performance of loess has been confirmed by various tests [39, 41, 42, 47, 49]. However, the tensile strength of fiber-reinforced loess has not raised enough attentions, although its impact on geohazards in loess areas is critical. For other types of soils, the significance of fiber content and length on the tensile strength has been concluded [10, 25, 37]. It requires further tests to optimize the fiber conditions before establishing the fiber reinforcement scheme for loess projects. Fiber reinforcement can change the failure morphology of soil, as captured by high-pixel camera [4, 14, 27]. The failure photographs could reflect the change of the plasticity of the soil before and after fiber reinforcement, but it is limited to qualitatively analysis. The digital image correlation (DIC) method could well solve this problem. DIC has been widely used in triaxial shear, bending, and clay ring cracking tests [8, 9, 11, 43]. DIC was also preliminarily applied in tensile tests. Divya et al. [10] investigated the effects of fiber content and fiber length on the tensile strain and crack formation of fiber-reinforced silty soil based on the DIC technique. Faghih Khorasani and Kabir [12] used the DIC technique to discuss the effect of short tire fiber on enhancing the tensile properties of adobe. However, since the test method for the surface strain field in uniaxial loading of loess is not yet refined, DIC-assisted uniaxial tensile tests are urgently needed to reveal the mechanical behavior of fiber-reinforced soils and further interpret the mechanism of the influence of fiber length and fiber content.
Due to the difficulty in assessing the tensile strength of soil, test methods such as split tensile tests, direct tensile test and horizontal tensile test were applied [2, 3, 29, 33, 37], but there is still no unified test standard. Considering the merits of uniaxial compression test in determining the uniaxial compressive strength of soil, the attention of some scholars has focused on the feasibility of using UCS to estimate UTS. Experiments have proved the proportional relationship between the tensile strength and the compressive strength of soil [31]. In terms of fiber-reinforced soil, Yilmaz [51] found that the ratio of splitting tensile strength (STS) to UCS of fiber-reinforced sand-clay mixture ranged from 0.17 to 0.31. Consoli et al. [3] reported that the ratio of STS to UCS of fiber-reinforced silt-lime mixture was 0.11–0.20. Festugato et al. [13] derived the relationship between STS and UCS of fiber-reinforced cemented soil based on the superposition concept of failure strength contribution and pointed out that the ratio of STS to UCS is a constant, which is 0.156. The insight of the above studies is that the tensile strength could be estimated by the compressive strength of soil. In addition to the relationship between UCS and UTS, it still necessitates to consider the effects of fiber length and fiber content when constructing the mathematical relationship for fiber-reinforced soil.
Currently, the performance of fiber-reinforced soil is mainly studied by laboratory unit tests, such as direct shear, unconfined compressive, and triaxial tests. These laboratory unit tests can only evaluate the reinforcement effect through the obtained macroscopic mechanical indexes, and cannot clarify the mesoscale mechanism of fiber-reinforced soil. Therefore, it is necessary to use numerical simulation methods to study the evolution of mesoscopic parameters and local failure characteristics during the loading process of fiber-reinforced soil. Among them, the discrete element method (DEM) is suitable for solving the problem of the heterogeneous structure of fiber-reinforced soil, which can not only obtain its macroscopic mechanical response but also record the contact force distribution between fiber and soil particles and displacement field during the loading process [18, 22, 35]. Ibraim et al. [19] investigated the evolution of pore ratio distribution, rearrangement of soil particles, and dynamic change of stresses in fiber-reinforced soils during compaction by DEM. Gong et al. [15] used DEM to simulate the microscopic mechanism of strength anisotropy of fiber-reinforced sand in direct shear tests. Yang et al. [50] studied the influence of different fiber orientations on the contact force distribution between fiber and sand particles. Therefore, it is meaningful to use DEM to simulate the tensile behavior of fiber-reinforced loess and to analyze the mesoscale mechanism of fibers to resist the tensile failure of loess.
In this study, the uniaxial tensile tests facilitated by the DIC technique were carried out on basalt fiber-reinforced loess to study the effects of fiber content and fiber length on the tensile strength, failure mode, and surface strain field. Then, a UTS prediction model was established based on the UCS of fiber-reinforced loess. Finally, the DEM package Particle Flow Code (PFC2D) was used to analyze the mesoscale mechanism of fiber-reinforced loess. The results will help to further reveal the mechanism of fiber reinforcement on the strength and deformation of loess, and to evaluate the relationship between tensile and compressive strength of fiber-reinforced loess.