Effects of dry-wet cycles on three-dimensional pore structure and permeability characteristics of granite residual soil using X-ray micro computed tomography

Ran An, Lingwei Kong, Xianwei Zhang, Chengsheng Li

a College of Urban Construction, Wuhan University of Science and Technology, Wuhan, 430065, China

b State Key Laboratory of Geohazard Prevention and Geoenvironment Protection, Chengdu University of Technology, Chengdu, 610059, China

c State Key Laboratory of Geomechanics and Geotechnical Engineering, Institute of Rock and Soil Mechanics,Chinese Academy of Sciences, Wuhan,430071,China

Keywords:Granite residual soil Dry-wet cycle X-ray micro computed tomography (micro-CT)Three-dimensional (3D) pore distribution Seepage simulations Permeability

A B S T R A C T

1. Introduction

Granite residual soil is a natural soil-like material,which widely exists in the foundation stratum of building and infrastructures in subtropical and tropical zones (Rahardjo et al., 2011; Zhai et al.,2016; An et al., 2021). Moreover, the landslides of granite residual soil (Coutinho et al., 2019) are challenging issues impacted by the engineering properties including water-swelling effects (Rao and Revanasiddappa, 2002; Kim and Kim, 2010), spatial variability(Branco et al., 2014), moisture susceptibility (Da Fonseca et al.,1997), and high probability to collapse (Saffari et al., 2020).Compared with general soil materials,granite residual soil exhibits significant anisotropy, and the macroscopic mechanical indicators are more complicated due to the effects of dry-wet environment(Liu et al.,2021).As a product derived from the in situ weathering and decomposition of granite,this type of material exhibits a special grain distribution from clay to gravel, as well as an intricate structure of primary pores and cracks (Mohamedzein and Aboud,2006; An et al., 2020). The microstructure, as an important intrinsic characteristic of soils, noticeably impacts the physical,hydraulic and mechanical behaviors of soils(Luo et al.,2010;Ye and Li, 2018). Due to seasonal climate alterations, the microstructural characteristics of granite residual soil are easily affected by periodic water infiltration and evaporation;thus,a path is created for water and ionic species to penetrate into soil particles,thereby leading to a variation of its permeability (Kim and Kim, 2010; Zhang et al.,2017). However, the damage of microstructures in soils attributed to dry-wet cycles is hard to characterize and quantify using traditional microstructural detection methods, such as mercury intrusion method and scanning electron microscope (SEM) (Tang et al.,2012; Park et al., 2015). Several studies have been conducted to assess the effects of dry-wet cycles on the physical and mechanical properties of residual soil. It has been reported that the particle connections, soil fabric, and porosity of granite residual soil vary noticeably. As a result, the mechanical properties and stability of granite residual soil are down-regulated while the permeability is up-regulated for the effects of multiple dry-wet cycles (Rao and Revanasiddappa,2006; Sayem et al., 2016; Kong et al., 2018).

Extensive studies have been conducted on the effects of dry-wet cycles on the macroscopic physical and mechanical properties of soils,whereas our insights into microstructure variations should be deepened in further studies (Otalvaro et al., 2016). A series of testing methods has been employed to analyze the pore distribution characteristics within soils, which covers visual observation(Pires et al.,2008),resistivity survey(Zeng et al.,2018),SEM(Wan et al., 2014), mercury injection porosimeter (MIP) (Zhang et al.,2016), nuclear magnetic resonance (NMR) (Mohnke and Yaramanci, 2008), and X-ray computed tomography (CT) (Grevers et al., 1989). X-ray CT has been commonly employed to explore the pore and crack structures of porous material for its several advantages, including its non-destructive process to samples, precise and quantitative detection,as well as 3D structure visualization(Ojeda-Maga et al.,2014).Several researchers preliminarily used CT tests to detect the internal structure of residual soil and to explore the changing patterns of its pore structure during water infiltration(Sun and Tang, 2019; Li et al., 2020). However, there are few relevant reports on the microstructure evolution of residual soil due to dry-wet cycling effects by using CT investigations.

Based on the X-ray micro computed tomography (micro-CT)scanning results of porous materials, some researchers further explored the permeability by seepage simulations.Berg et al.(2016)conducted a series of micro-CT scans for investigating the permeability in sandstone samples and have obtained a certain progress.Starnoni et al. (2017) used 3D CT scanning results to establish the numerical model of Navier-Stokes flows in the pores of a porous rock and evaluated the seepage flow capacity of rock based on the seepage simulations. Wang et al. (2019) focused on the seepage mechanism in six coal samples and concluded that the permeability of coal increases as the degree of pore abundance increases.Despite the developments of seepage simulations based on micro-CT results, very few researches were performed on the comprehensive understanding of 3D pore structures and permeability in soils using the micro-CT results.

The present study aimed to determine the characteristics of pore structure in granite residual soil impacted by different numbers of dry-wet cycles using micro-CT scans. Given the scanning results,the volume content of pores, pore size distributions and the CT reconstruction images of pore structures were characterized and their variation patterns were analyzed. This study will provide more insights into the micro to macro physical variations of granite residual soil under the influence of multiple dry-wet cycles.

2. Materials and methods

2.1. Material properties

The soil samples tested here were obtained from a foundation pit slope (4-6 m depth) in Xiamen Island (longitude 118°04′E,latitude 24°26′N),in southeast coastal areas of China.Through the excavation in weathered granite during the metro line construction from November 2014 to August 2016, the foundation pit was formed. In the course of excavation, the soil in slope surfaces underwent the change between dry and rainy seasons for multiple times. As a result, the dry-wet effect of granite residual soil under seasonal climate alteration is critical during the formation of the foundation pit with weathered soils (Ng and Leung, 2012).

The physical properties of granite residual soil, including initial moisture content(w),dry density(ρd),void ratio(e),specific gravity(Gs) and Atterberg limit (IlandIp) were measured according to the Chinese national standard GB/T 50123-1999(1999).Table 1 outlines mineral compositions and basic physical properties of the granite residual soil.According to the analysis of the X-ray diffraction(XRD)referring to ASTM D4926-15 (2015), the granite residual soil primarily consists of non-clay minerals (e.g. quartz and hematite) and clay minerals (e.g. kaolinite and illite). Due to the incomplete weathering of parent rocks, the granite residual soil covers a considerable number of coarse grains that are mainly composed of quartz (An et al., 2021). The grain size distribution curve was measured according to ASTM D422-63 (2002). As shown in Fig.1,the granite residual soil studied in this paper covers high contents of clay particles,as well as gravel and sand particles, which acts as an intermediate geomaterial between clayey and sandy soil.

Table 1 Basic physical and chemistry parameters of granite residual soil.

Fig.1. Grain size distribution curve of granite residual soil.

2.2. Sample preparation

The exploratory trench sampling method was used in this study.Undisturbed samples of granite residual soil with nature moisture contents were collected at the foundation pit slope. Cubic soil samples wrapped with plastic films on their surface were elaborately cutted from the slope. Then, the soil samples were placed carefully in wooden boxes filled with shock absorbing materials and transported discreetly to the laboratory.Cylinder samples with a size of 80 mm in height and 38 mm in diameter were prepared from the large samples (Fig. 2a) referring to Chinese nation code GB/T50123-1999 (1999). To ensure the smoothness of the subsequent procedures, the surfaces of samples were trimmed and polished,as shown in Fig.2b.Given the climate of Xiamen Island,the mean monthly temperature shows a variation from 11.2°C in January to 34.7°C in July, and the mean monthly precipitation varies from 26.1 mm in December to 215.2 mm in July.

In the present study, the tested samples underwent 0,1, 2, 3, 5 and 8 dry-wet cycles. For each cycle, water absorption was performed using the vacuum saturation method, and a temperaturehumidity controlling chamber (Fig. 2c) was adopted to simulate the desiccation environment based on weather data. First, the sample was air-dried up to the soil shrinkage limit (nearly 12%)with temperature and relative humidity of 40°C and 75%for about 48 h,respectively.Next,the tested sample was saturated for about 24 h using a vacuum saturator(Fig. 2d)and then submerged at an ambient temperature of 20°C for about 48 h. Finally, the sample was air-dried in the chamber until the water content reached the natural moisture content(about 26%) for next procedures.

Fig. 2. Process of the tested sample treatment: (a) Field sampling, (b) Sample preparation, (c) Drying treatment, and (d) Wetting treatment.

2.3. Procedure of micro-CT scans

The micro-CT test instrument, tested sample with the size of 38 mm in diameter and 80 mm in height, and original results are presented in Fig. 3. To investigate the microstructural characteristics of granite residual soil that experienced different dry-wet cycles, the Zeiss Xradia-410 micro-CT scanner with an accuracy of 0.1 μm was adopted. The micro-CT scanning system primarily consists of the X-ray emission source,the X-ray detector,a rotating table and a computer system that retrieves image information.During the experiment,the sample was placed on the rotating table for a panoramic scan at the degree of 0°-360°.Subsequently,the Xray emission source emitted rays to penetrate the sample,and the components within the soils were detected with the X-ray detector.The sample height exposed to the X-ray was nearly 40 mm, while the diameter of the scanned sample was approximately 30 mm.After each scan,1320 grayscale images with a spatial resolution of 15 μm of the tested sample were obtained.

Fig. 4. 3D reconstruction of digital soil model based on micro-CT images: (a) 2D grayscale image, (b) 2D rendering image, (c) 3D model of soil, and (d) 3D models of different components.

The X-ray CT technique refers to a digitalized image reconstruction from a series of grayscale images, reflecting materials with a range of densities and displayed as 3D models.When X-ray penetrates a tested sample,differences will exist in the attenuation coefficients for X-ray intensity of the materials with various densities, which is reflected as different grayscales within CT images.The attenuation coefficients(I)of various materials is calculated by the following equation (Wang et al., 2019):

whereI0andIdenotes the intensities before and after the X-ray penetrates the object, respectively; χ indicates the penetration length of incident X-rays;ρiis the density of probed material(Braz et al., 2001); and μmrefers to the absorption coefficient of unit mass.

On the whole,μmand χ are only related to the wavelength and emission distance of testing X-ray. In a specific CT scan test, the attenuation coefficient (I) displays a negative relation with the density in an exponential manner, suggesting that the image brightness is down-regulated with the rise in the material density.Accordingly,materials with a range of densities displayed different grayscales in CT images(Van Stappen et al.,2019).

According to a two-dimensional (2D) CT image, it was found that the micro-CT scanning technique reflected the real structures of the tested sample precisely,which was truncated horizontally for graphics display(Garbout et al.,2013).Based on the results of XRD tests in this study,the granite residual soil consists of clay minerals,quartz,hematite(iron-bearing mineral)and pores among the solid mediums.Since there are noticeable differences in the densities of media, the hematite, quartz, clay and pores in the soil can be discerned by micro-CT images.As shown in Fig.3,the hematite,quartz particles, clay and pores in the granite residual soil are denoted as white,light gray,dark gray,and black areas in the primary grayscale images of the tested sample, respectively.

3. Results and discussion

3.1. 3D reconstruction model of soils

The reconstruction and processing procedures of the 2D image and 3D digital model are illustrated in Fig. 4. First, the original 2D CT grayscale images are marked by different colors. As shown in Fig.4a,the hematite,clay,quartz,and pores are marked by yellow,gray, red and green, respectively. Though the 2D CT image is capable of reflecting the internal structure of the sample at a specific height of the tested sample, the stereo display is insufficient,which is of high implication to the microstructure analysis.Accordingly,the 2D images are superimposed to rebuild a 3D digital model of the real granite residual soil, as presented in Fig. 4c. For deeply investigation,a range of components in soils was extracted to display the 3D models of the hematite, quartz, clay, and pores,respectively, as given in Fig. 4d. By harvesting several contiguous cross-sectional images from CT scans,complete data of the 3D pore volumes in soil samples can be achieved.

Fig. 5. The 3D pore model of granite residual soil during the dry-wet cycles.

For clear presentation of the propagation path of pores in granite residual soil, 3D pore models under dry-wet cycles were reconstructed.The color rendered geometry in the cubic box with a side length of 20 mm represents the pores(Fig.5).The pore space within the observed sample is gradually enlarged with the rise in dry-wet cycles.In the internal structure of the granite residual soil without drying and wetting treatments, the size of pores and the degree of connectivity between pores remain to be small and low,respectively. After the first dry-wet cycle, the pores and cracks in soils expanded obviously, and the number and size of connected cracks increased. After 3 dry-wet cycles, an intricately connected crack network was formed in the granite residual soil.After 5 drywet cycles, the pores continued to fill the internal space of the granite residual soil,and the connectivity of the pore structure was significantly enhanced. It is therefore speculated that due to the damage effect of periodic infiltration and precipitation of water molecules in the soil matrix, numerous shrinkage cracks occur. To quantitatively study the progressive damage of the sample, the quantified pore distribution features,e.g.the volume contents and size distributions of pores, are analyzed based on the mentioned four separated 3D pore models.

3.2. Pore-volume distributions

Numerous studies have been conducted considering soil a porous medium with considerable pores and cracks, of which the pore content and connectivity can be calculated using the 3D digital model based on CT scans (Monga et al., 2007; Li et al., 2012; Sun et al., 2012; Gao et al., 2019). Using the 3D CT digital model, a pixel volume in a certain substance refers to the total volume of pixels that are covered in the material.The AVIZO 9.0 software can create a powerful platform for micro-CT result analysis, which can be used to effectively achieve 3D structure visualization (Zhang et al., 2018). With the use of the software, the pore volume and the total volume of the scanned sample were calculated based on voxel counting; subsequently, the porosity (φ) as well as connectivity(η)of the granite residual soil was calculated by Eq.(2).In this study, the porosity refers to the pores with relatively large sizes,which can be detected by micro-CT tests.The results of porosity and connectivity obtained by micro-CT tests are given in Table 2.

Table 2 The pore parameters calculated from micro-CT tests.

where vporerefers to the total pixel volume of pores and cracks,vsoilis the total pixel volume of soil sample, and vcon-porerefers to the total pixel volume of connected cracks.

Fig. 6a shows the porosity with increasing dry-wet cycles. According to the regression analysis of cycles and pore distribution parameters,the porosity of the tested sample increases nonlinearly with the number of dry-wet cycles,in form of a specific exponential function. Besides, the connectivity increases linearly within a limited number of dry-wet cycles according to Fig. 6b.

Fig. 6. Pore parameters of the sample under 0-8 dry-wet cycles: (a) Porosity, and (b)Connectivity.

The probability distribution curves of the pore volume of the 3D pore model are plotted in Fig.7.We calculated the number of voxels contained in the pore phase and the integral soil sample from the 3D models.Then,the pore volumes were converted from the voxel number of pores according to the size of a voxel.It suggests that the distribution range of pore volume in the sample is 0.01-100 mm3,which is wide. As revealed by the variation of the pore volume distribution curve, the peak pore volume of granite residual soil rises while the proportion of peak pore volume declines with the increasing dry-wet cycles. The peak pore volume rises from 0.011 mm3to 0.0125 mm3,and the proportion of peak pore volume decreases from 28.32% to 17.29%. Generally, the pore structures of the granite residual soil are dramatically altered under periodic dry-wet cycles.

Fig.7. Pore size distribution curves of the sample under different dry-wet cycles(DW:dry-wet).

Chi-square (χ2) distribution is a probability distribution commonly used in probability theory and statistics (Lipiec et al.,1998). According to the characteristics of pore-volume distributions, χ2distribution function is substituted to express the porevolume distribution curves of granite residual soil, as displayed in Fig. 8. The χ2distribution function refers to a random variable continuous probability function with a fitting parameter,named as degree of freedom (n).The χ2probability distribution is flexible to use its function for data fitting in the expression of the pore-volume distribution curve(Hatano et al.,1988).During the dry-wet cycles,cracks in the soil sample continuously expand, which can be quantitatively described by adjusting the parametern. The mathematical formula of χ2probability density function is written as

Fig. 8. Pore size distribution curves of the sample under 0-8 dry-wet cycles: (a) 0 cycle, (b) 1st cycle, (c) 2nd cycle, (d) 3rd cycle, (e) 5th cycle, and (f) 8th cycle.

where ε denotes the volumes of pores.

Fig. 8 shows the pore volume distribution with a range of drywet cycles and the corresponding fitting curve of χ2probability function. According to the results, the χ2probability distribution function can be adopted to approximate the pore-volume distribution curve under different dry-wet cycles by altering the fitting parametern. The χ2distribution function is capable of effectively expressing the probability distribution patterns of pores in granite residual soil during the dry-wet cycles, and it also reveals the evolution patterns of pore structures.

3.3. Pores classification based on volumes

The image-derived pore volumes were calculated considering all the pores in the soil. According to the research of Pires et al.(2020), the 3D pore volume distributions in soil were determined based on the pores classified in different logarithmic volume intervals: < 0.01 mm3, 0.01-0.1 mm3, 0.1-1 mm3, and >1 mm3,which were named as micropore, mesopores, macropores and cracks, respectively. The proportional variations of the pores with different volume ranges are presented in Fig. 9. As the number of dry-wet cycles increases,the proportion of pores with volumes less than 0.01 mm3declines, while the proportion of pores with volumes larger than 0.1 mm3increases distinctly (especially for the pores with volumes greater than 1 mm3). Meanwhile, the proportion of pores with volumes ranging from 0.01 mm3to 0.1 mm3remains nearly unchanged.

Fig. 9. The proportion of pores with different volume ranges.

The volumetric changes of pores in different types are plotted in Fig.10.The decreasing amplitude of micropore’s proportion is most noticeable, and the increasing amplitude of crack’s proportion exhibits the highest significance,followed by the macropore,while the varying amplitude of the mesopore’s proportion has the smallest variation.According to the results of pore-volume probability distributions,considerable primary pores that display small volumes in the granite residual soil gradually expand and connect the cracks exhibiting large volume and high connectivity during the multiple dry-wet cycles.The variation of the pore volume probability distribution is the most remarkable in the first dry-wet cycle,and decreases with subsequent cycles.Due to dry-wet cycles,the hematite and quartz particles are almost non-deformed while the deformation of clay is significant, tending to enlarge pores in the clay matrix. The pore structures appear to be rigid progressively,thus the pore structures in granite residualsoilare hard to deformfor furtherdry-wetcycles.This phenomenon is attributed to the fact that the tensile stress among the soil particles is gradually weakened with the gradual dissipation of matric suction during dry-wet cycles.When the tensile stress among soil particles is completely below the threshold of cracking initiation,the network of cracks enters a stable stage.

Fig. 10. The variation amplitude of different types of pores under different dry-wet cycles.

3.4. Permeability evaluation based on seepage simulations

As one of the most significant hydraulic parameters of geomaterials,permeability refers to the property of water permeating through pores among soil particles (Li et al., 2009). In the present study,numerical simulation of the characteristics of multi-channel seepage was performed for low-pressure water in a 3D model of pore structure in the samples of granite residual soil using AVIZO.The software provides a function(AVIZO XLab Hydro Extension)to conduct absolute permeability testing simulation based on the Navier-Stokes equations, in which standard turbulence was adopted and the fluid flow was set as water in normal room temperature.By using 3D pore models for seepage simulations,it is not necessary to mesh the digital model of soil samples as finite element method needs, which allows the results of absolute permeability and streamline models calculated by seepage simulations to reflect the accurate permeability characteristics of granite residual soil.

In this study, the Navier-Stokes equations of incompressible fluid flows are used to simulate the seepage process in pore structure.The governing equations for the seepage simulation in a given soil based on computational fluid dynamics(CFD)methods of AVIZO software are presented as (Starnoni et al., 2017):

whereuis the flow velocity,tis the time,ρ is the density of fluid,Eis the internal energy,σ is the stress tensor,qis the heat flux, and ∇denotes the Laplace operator.

Based on the Navier-Stokes equations, the absolute permeability (k) of soil is calculated by(Romano,1993)

whereQis the volumetric flow(m3/s);μ denotes the viscosity coefficient of fluid,which is 1.0087×10-3Pa s at temperature of 20°C;Ldenotes the length of the calculated sample,andL=0.02 m;A is the cross-sectional area of flow,andA=0.004 m2;and ΔPdenotes the fluid pressure difference between import and export values,and ΔP= 1 MPa.

In geotechnical engineering, the hydraulic conductivity (K)is usually used to describe the permeability of soils, which can be converted from the absolute permeability (k):

where γwdenotes the unit weight of water(9.8 kN/m3)in this case.

The 3D models of streamlines of granite residual soil under different dry-wet cycles are shown in Fig.11.The complete seepage flow path (fluid flows from top to bottom of the model uninterruptedly)observed in the numerical model of soils without dry-wet cycle is the least, and the vast majority of the seepage streamlines are cut off halfway.As the number of dry-wet cycles increases,the number and distribution density of streamlines in seepage simulations continue to increase. The phenomenon interprets that the damage effect of dry-wet cycles makes the quantity, shape and connectivity degree of connected cracks in the granite residual soil vary. Besides, the results of absolute permeability (k) and the corresponding hydraulic conductivity (K) calculated by the seepage simulations are presented in Table 3, which shows the absolute permeability as well as the hydraulic conductivity increases gradually with the increasing number of dry-wet cycles.

Fig.11. Velocity streamlines of residual soil: (a) 0th cycle, (b) 1st cycle, (c) 2nd cycle, (d) 3rd cycle, (e) 5th cycle, and (f) 8th cycle.

Table 3 The calculated results from seepage simulation.

According to the regression analysis,it is found that the hydraulic conductivity of soil increases exponentially with the porosity,while linearlywiththeconnectivity,asshowninFig.12.Differentfromsandy soils,the granite residual soil is often regarded as an intermediate soil which exhibits intricate grain gradation and pore distribution characteristics.Itspermeabilitybehaviorismorecomplicated,whichisnot only related to the porosity,but also closely related to the pore connectivity,morphology,amountandotherfactors.Theconnectedcracks are the effective paths for seepage formation,therefore,the volume content and connectivity of pores are directly related to the permeability.Fromtheresultsofseepagesimulations,itcanbeobservedthat the pore connectivity has an important influence on permeability of the soil sample. Due to dry-wet cycles, the clay minerals such as kaolinite and illite have strong hydrophilic characteristics and easily expand when touching water. The subsequent chemical eluviation reaction may also change the mineral crystal morphology and pore structures(Hironoetal.,2003).Sometinyporesinparticlesweremore developed into connected cracks where water flow passes through under the dry state,which led to the increase in the permeability coefficient accordingly.As the number of dry-wet cycles increases,the connected cracks in granite residual soil are enlarged and expanded continuously. Thus, the number of seepage channel of water in soil increasesgradually,aswellasthepermeability.Eventually,thestability of seepage field in the foundation is affected due to the increase of soil permeability caused by the dry-wet cycle,which may induce correspondingengineeringdisasters(YoshidaandHallett,2008).Therefore,it is of critical importance to concern the behavior of formation and expansion of connected cracks in foundation soils.

Fig. 12. The relationship between hydraulic conductivity and pore parameters: (a)Porosity, and (b) Connectivity.

The seepage simulation method makes full use of the results of micro-CT scans,although the simulating conditions are not perfectly closer to those of the actual situation. To verify the numerical simulation results, the constant-water head penetration experiments were performed on the cylinder samples with different drywet cycles according to ASTM D2435-03 (2003). The measured hydraulic conductivity (Km) is listed in Table 4. The relative errors may result from the poor porosity and connectivity of these samples such that the calculation of cumulative infiltrations during seepage simulations become less accurate.Accordingly,the calculation error of the original sample with low cracking connectivity is relatively high,and declines with the increasing number of dry-wet cycles.In general, the calculated values of hydraulic conductivity are comparable with measured ones with an acceptable error margin,indicating that the seepage simulation method based on micro-CT scanning results is feasible to estimate the permeability characteristics of soil.Besides,the seepage simulation results provide a proof for the high accuracy of the micro-CT scans.

Table 4 The calculated and measured results of seepage simulation.

It is worth mentioning that the pore characteristics of different soils generally have great differences.Granite residual soil is widely considered as a special weathered soil.Its macroscopic mechanical properties and microstructure are different from those of normal sedimentary soil due to its special formation history and structural characteristics.The results of seepage simulations in this study are based on the assumption of cracks with upper and lower connectivities in the scanned soil sample. Further study is needed to determine whether this simulation method is applicable for other soils.Meanwhile,the attempt of micro-CT scans and seepage simulations in this paper can provide a new perspective for the permeability research on other fissured soil types.

4. Conclusions

A series of micro-CT scans has been performed for mesotomography observations of the granite residual soil subjected to multiple dry-wet cycles. Subsequently, 3D reconstruction and seepage simulation based on CT images were further analyzed.The main conclusions are as follows:

(1) Based on micro-CT results, the morphology of the mesostructure composed of hematite, quartz, clay and pores in granite residual soil has been proven to be different from that of the general soil materials. Moreover, variations of pore structures observed and analyzed by the 3D reconstruction technique are the result of dry-wet treatments. The specific variation is attributed to the augment of the volume and connectivity of pores in granite residual soil as the number of dry-wet cycle increases.

(2) The values of porosity and connectivity are positively correlated with the number of dry-wet cycles, expressed by exponential and linear functions,respectively.The micro-CT scanning results indicate that small pores in the original sample appear to expand and connect after drying and wetting treatments.

(3) The pore volume probability distribution curves of granite residual soil coincide with the χ2distributions,verifying the assumption of the χ2probability distribution. Furthermore,the pore-volume distribution histogram reveals that the volume of cracks and macropores increases while that of micropores decreases. Under the action of dry-wet cycles,considerable small pores are gradually transformed into cracks with large volume and high connectivity.

(4) The amount of streamlines in seepage simulations continues to increase, indicating that periodic drying and wetting treatments change the quantity,shape and connectivity degree of cracks. Correlations between pore parameters and hydraulic conductivity confirm that the dry-wet cycles affect the permeability in terms of porosity and connectivity.The variation of pore structures is the intrinsic reason for the difference in the hydraulic conductivity of soil subjected to different drywet cycles.

(5) The calculated hydraulic conductivity is comparable with measured ones with an acceptable error margin in general,indicating that the seepage simulation method based on micro-CT results is feasible to estimate the permeability characteristics of the granite residual soil.

Declaration of competing interest

The authors declare that they have no known competing financial interests or personal relationships that could have appeared to influence the work reported in this paper.

Acknowledgements

The present study is supported by the National Natural Science Foundation of China(Grant Nos.12102312 and 41372314)and State Key Laboratory of Geohazard Prevention and Geoenvironment Protection Open Foundation, Chengdu University of Technology,China (Grant No. SKLGP2021K011).