Combination seepage test study of sandy soil and the filter Open Access

Dams are important structures for the development and utilization of water resources. According to statistical data, 3,539 dams were damaged in China from 1954 to 2018 (Zhou et al. 2020). Failure occurring in earth–rock-filled dams represents up to 95% of all dam failures. Most of those failures pose a significant threat to human life and property safety. A detailed study found that 66% of earth–rock-filled dams were damaged by dam seepage (Calamak & Melih Yanmaz 2018). Sherard (1984) clearly stated that the filter layer forms the first line of defense against seepage damage in earth–rock dams. The risk of permeation deformation and damage to the impermeable body can be effectively reduced by installing a filter layer between the soil impermeable body and dam filling materials. As soil material for the filter layer, sand or gravel with permeability coefficients exceeding the permeability coefficient of the impermeable soil material are usually selected. The main role of this filter layer can be summarized as ‘protection of soil particles and reduction of hydraulic gradients’ (Liu & Xie 2017). This not only ensures that the impermeable body and the particles at the dam base will not be washed out by water because of water pressure, but also, water pressure will be completely or mostly dissipated after the seepage water enters the filter layer (Liu & Xie 2017; Yang et al. 2019). Moreover, the filter layer can coordinate the deformation between the impermeable body and the dam shell to prevent excessive uneven settlement cracks in the dam (Kim et al. 2022; Kumar et al. 2022). These unfavorable conditions can be avoided when a reasonable filter layer is designed.

Based on engineering practices, Terzaghi proposed both the theory of designing a filter layer to prevent soil seepage damage and filter layer design guidelines, providing theoretical guidance for practical projects (Kumar et al. 2022; Wang et al. 2022). The role of protective soil particles is determined by the coarsest particle composition allowed for the filter. In other words, the average diameter of the pores in the filter layer is smaller than the control particle size of the protected soil. The control particle size is the particle size in the soil that has a significant effect on seepage damage. The reduction of hydraulic gradients determines the finest particle composition allowed for filter layer materials. The filter material will have a stronger water drainage capacity and a better effect of reducing the hydraulic gradients when the filter layer has a large permeability coefficient; thus, particle gradation is an important factor for determining the effect of the filter material (Wang et al. 2013; Vakili et al. 2018; Caldeira 2019; Yang et al. 2019). Alam (2021) studied the seepage characteristics through numerical analysis in an earth dam, using different filter materials such as sand, gravel, and rock. Analysis showed that changes in seepage velocity, hydraulic gradient, and discharge are based on the coefficient of permeability of materials that were used in this filter layer.

Many scholars have conducted extensive research and have proposed corresponding design guidelines for the filter material. The range of filter grades for a specific grade of impermeable soil material can be identified by calculation. However, the combination impermeability performance regulation of impermeable material and filter material needs to be studied further, especially when the particle composition of the filter material varies within the gradation range. In the design of the filter layer, the thickness of the filter layer is mainly determined by experience and actual conditions. The filter layer could not meet the required function of drainage water, reduction pressure, and coordination deformation if the filter layer is too thin. In contrast, if the filter layer is too thick, the cost and emerging waste will increase. Shi et al. (2020) analyzed the effect of changes in the thickness of the filter layer on the seepage characteristics of the dam using numerical simulations (Zou et al. 2013; Zedan et al. 2017; Shi et al. 2020). Shirazi et al. (2021) used SEEP/W software to study the dam seepage characteristics under the effect of anti-seepage systems and filter layers with varying lengths, angles, and locations. The results indicated that with filter expansion, the seepage flow will decrease. Most design guidelines for filters focus on controlling the gradation of soil materials used, and compaction is an important quality control index in the actual engineering process of soil filling. Through a combination of indoor tests and theoretical calculations, To et al. (2020) found that its relative density significantly affects the effect of the filter layer and the pore diameter of the filter layer. Therefore, the influence regulation of the relative density change of both impermeable material and anti-filtration material on their combination impermeability characteristics needs to be studied in depth.

In this study, a natural graded soil material was selected as impermeable material, and filter gradation characteristics were obtained by calculation. This combination of impermeable material and filter material impermeability test was employed to study the effect of soil material particle composition, relative density, and soil layer thickness on the resulting combination impermeability characteristics. This study presents several interesting analyses, such as the effect of variations in the gradation of the two layers of soil materials within their design range on the seepage results. Knowing the relationship between variations in filter layer thickness and seepage characteristics enables the designer to become more confident in the choice of thickness. The relative density in the experimental study and the quality control of the actual engineering construction correspond. The research results provide a reference for the scientific design of filter layer and the effective impermeability improvement of the impermeable layer.

The upper and lower limits of filter material gradation are shown in Grading 3 and Grading 4, both of which were identified as well graded gravel (GW). The maximum and minimum dry densities of the filter material were 2.07 g/cm³ and 1.80 g/cm³, corresponding to pore ratios of 0.30 and 0.50, respectively. During the test, the corresponding graded soil material was configured according to the curve.

Before starting the test, the required soil material was prepared according to the test plan. To prevent water flow from the side walls and corners of the apparatus (which would affect the test results), before loading the sample using glass cement, the corners of the specimen barrel were treated and the inside of the instrument was evenly coated with Vaseline (GB/T 50123 2019). Each layer of soil sample was compacted to the desired relative density. To ensure close contact between each layer of soil material, scraping was conducted between each layer of soil material. The specimens were saturated by a slight gradient of water pressure after the filling had been completed. At the beginning of the test, the water head was raised until water came out of the outlet; then, the head of the pressure measurement and the outlet flow were read.

During the test, readings were taken at 30 min intervals after raising the hydraulic gradient. Each hydraulic gradient was measured at least twice, and the average value was taken. If the specimen does not show any changes (e.g., because the velocity of permeability does not increase with time, the seepage pressure does not change, or fine soil particles do not move), the next level of the hydraulic gradient was tested. If fine particles are found to run into the filter layer through the contact surface, or if the hydraulic gradient in the filter layer is found to have increased, then this level of head and the subsequent steps in each level of head test duration was extended to 2–4 h. The test was ended when the soil sample was raised, the flow rate continued to increase, or the hydraulic gradient in the protected soil decreased, or the hydraulic gradient in the filter layer was equal to the hydraulic gradient in the protected soil.

To study the protection effect of the filter material on the impermeable material combination seepage tests were conducted. The influence of grading, relative density, and thickness of the impermeable and filter material on the seepage failure resistance was studied. In total, eight tests were conducted in which different combinations were selected. The specific test programs are shown in Table 1.

The combination seepage test results of different gradation, relative density, and thickness combinations of impermeable materials and filter materials obtained through indoor tests are shown in Table 2. All eight combination seepage tests ended with soil flow. The results of seepage tests with different combinations are clearly different. The permeability coefficients of impermeable materials in Tests 5 and 6 were significantly higher than those of other tests. The critical hydraulic gradient and the critical failure hydraulic gradient of the impermeable material in Test 8 were significantly higher than those of other tests.

Figure 6 shows the seepage velocity curves of specimen over time. In the case of the same level of hydraulic gradient, the variation of seepage velocity is very apparent at the initial stage, and diminishes as the test proceeds. However, the time required to reach stability is longer, and the trend of seepage stability does not emerge in specific hydraulic gradients. This phenomenon can also be found in the results of Kim et al. (2022). The reason for this phenomenon is that with increasing hydraulic gradient, water flow exerts greater pressure on the fine particles in the soil sample. This pressure causes the fine particles to move, and this movement of the fine particles changes the seepage path, leading to the phenomenon of changing seepage velocity. Because of the action of water flow on soil particles, some of the fine soil particles move, thus consuming energy to work on soil particles. This consumption of water flow energy thus leads to a decrease of its flow rate. The larger the hydraulic gradient, the greater the water pressure it generates on soil particles, which can cause more and larger soil particles to move. The distance they move will also increase, thus increasing energy consumption. The more the water flow rate decreases, the longer it takes for the seepage rate to stabilize. In addition, as the fine-grained soil moves into the filter layer, soil particles remain in the filter near the contact surface of the two layers of soil. Under the filtering action of the filter material, a secondary filter layer forms (Wang et al. 2013; Correia dos Santos et al. 2015; Azirou et al. 2018; Yang et al. 2019). This further enhanced the filtration effect, increased the seepage resistance, and reduced the seepage velocity. The apparent variation of the critical gradient seepage velocity reached in Tests 2 and 4 were because Grading 2 soil is a piping type soil, where fine particles are more likely to move, thus affecting the permeability coefficient of the soil sample. In Tests 3 and 4, the filter material has a coarser gradation; therefore, the seepage velocity variation range is small, but variations happen more frequently. This phenomenon is related to coarser particles forming larger seepage pores.

The filter material gradation curves after Tests 1–4 are shown in Figure 7. The change of particle gradation was small after the test with fine filter material (Gradation 3), but the change of particle gradation was large after the test with coarse filter material (Gradation 4). When the grade of the filter material is fine, the pores in the soil sample are smaller, and the movement of fine particles is restricted to a certain extent under the action of filtering. This restriction also shows in the lack of change of the filter grade before and after Tests 1 and 2. The gradation of the secondary filter layer formed outside the control line of the filter gradation range, which is not beneficial to the drainage and pressure-reducing effect of the filter. However, with a coarser filter, more fine particles move from the impermeable material to the filter under the action of water flow. The filter grade changes more clearly after the test, as shown in the curves of Tests 3 and 4. Even if the filter grade is coarse, it still falls within the control line after the test under the action of seepage, which satisfies the corresponding effect.

Combined with the particle composition characteristics of the filter materials used in this study, the controlling constriction size of different filter material tests was calculated by the above equations. The controlling constriction size was 0.417 mm for Tests 1 and 2. The controlling constriction size was 1.167 mm for Tests 3–5 and Tests 7–8. For Test 6, the controlling constriction size was 1.303 mm. The d₈₅ of impermeable material Grading 1 and Grading 2 were 1.25 mm and 2.05 mm, respectively.

As shown in Figure 11, the values of different tests were all below 1, and thus, the soil particles in the impermeable material cannot pass through the control pore size of the filter. This also confirms that no fine particles were found to pass through the filter layer during the test, corroborating the better filtering effect of this filter. From Tests 1–4, the critical hydraulic gradient of specimens decreased significantly, regardless of whether the grading of the filter material or the impermeable material became coarser. In Tests 4, 5, and 6, the relative density of soil samples decreased or the grade became coarser, the value increased, and the critical hydraulic gradient of the specimen decreased significantly, under the same grade of the impermeable material. The decreasing relative density of the filter material caused the failure hydraulic gradient of the specimen to decrease significantly; therefore, in actual engineering applications, the dry density or relative density should be strictly controlled during filling.

In summary, setting reasonable filter material grading, thickness, and relative density is essential to the safety and stability of the resulting dam. Dams allow for the rational allocation of water resources to meet the needs of the population.

In this study, combination seepage tests of sand and gravel soil were conducted using different combinations of impermeable and filter materials. The experiments considered the effects of particle composition, relative density, and thickness of both impermeable material and filter material. The main conclusions are summarized in the following:

• (1)

  Within design gradings, changes of the grade of soil materials had a significant effect on the combination seepage test and the critical hydraulic gradient. The permeability coefficient decreased by 34% when the soil material grade changed from fine to coarse within the design grade. However, if the secondary filter layer consists of overly fine filter material, its drainage and pressure-reducing effects will be adversely affected.

• (2)

  The critical hydraulic gradient and failure gradient of the specimen decreased by 17 and 20% when the relative density of the filter material decreased from 0.7 to 0.5, respectively. Therefore, the dry density or relative density should be strictly controlled in actual engineering applications.

• (3)

  Increasing the thickness of the filter layer provides the soil with a greater critical hydraulic gradient, but also increases the cost. Therefore, this measure should be rationally applied in practice.

• (4)

  The controlling constriction size can reasonably describe the effect of changes in particle composition and relative density of the filter layer on the characteristics of combination seepage. The critical and failure hydraulic gradients decrease with increasing controlling constriction size. However, the controlling constriction size cannot be used to describe variations in filter layer thickness.

This work had no funding.

All relevant data are included in the paper or its Supplementary Information.

The authors declare there is no conflict.