Structural, soil-water, and dynamic characteristics of clay with different moisture and temperature histories

Variations in the hydro-mechanical properties of pavement subgrade soils are complex under long-term, repeated, and interlaced moisture and temperature (M-T) effects, such as wetting-drying (WD) and freeze-thaw (FT) processes. This study compares the soil structure, soil-water characteristic curves, and accumulated plastic strain (ɛ p) of compacted Heilongjiang clay with three different M-T histories. Experimental results demonstrated that (i) after M-T actions, structural pores develop while textural pores shrink, leading to a reduction in the water retention capacity and an increase in the desaturation rate. Following the stabilization of the M-T effects, the soil structure and soil-water characteristics of specimens with different M-T histories become similar; (ii) at high moisture content (w), ɛ p is more sensitive to moisture changes (including w and suction, s). Following the FT cycles, ε p and ε pa (the difference between the ε p at 5000th and 3000th cyclic loading) become more sensitive to moisture changes; (iii) after M-T effects, the ε pa-w relationship is nonlinear while the ε pa-logs relationship is linear. Different M-T histories generated differences in only the ε p at the beginning of the FT processes while such differences disappear after M-T effects have stabilized. The experimental data presented in this paper could prove valuable in understanding the behaviors of pavement soils under complex environmental conditions.


Introduction
The dynamic responses of subgrade soils under traffic loading determine the thickness and service life of pavement structures.Accumulated plastic strain (p) is a key dynamic response parameter of pavement soils, which continuously and significantly changes under vehicle loads and long-term and interlaced moisture-temperature (M-T) effects (such as wetting-drying, IOP Publishing doi:10.1088/1755-1315/1330/1/012064 2 WD and freeze-thaw, FT processes) [1].Consequently, it is vital to reasonably understand and predict the p considering the combined effects of traffic and environmental loads.
Pavement subgrade soils typically exhibit characteristics of unsaturated soils.Numerous theoretical and experimental studies have demonstrated that unsaturated soil behaviors are highly dependent on soil structure and soil-water characteristics, which evolve remarkably upon the impacts of M-T effects.Observations of several recent studies on the influences of M-T effects on the microstructure of soils, soil-water characteristics, and εp are summarized as follows: (i) FT cycles result in a gradual decrease in textural pores, an increase in structural pores, and generation of fissures in soils.Following WD cycles, textural pores increase, and structural pores decrease [2]; (ii) the rates of saturation and desaturation (i.e. the slope of the soil water characteristic curve) significantly change after the WD and FT cycles [3]; (iii) the variations in εp generally stabilize after three to five FT cycles [4]; (iv) during WD processes, the εp increases when suction decreases, moisture content increases and the number of WD cycles increases [6].
Existing studies mainly focus on the effects of independent FT or WD cycles, or FT and WD cycles that take place following predefined orders.However, pavement subgrade soils naturally undergo interlaced and random WD and FT processes.It is therefore uncertain whether the soil behaviors determined from the predetermined and simplified M-T histories are comparable and applicable to filed soils under more complicated and random M-T actions.
This study reports a series of experimental efforts that were devoted to understanding the influences of different moisture and temperature histories on the structural, soil-water, and dynamic characteristics of clays.Filter paper tests, mercury intrusion porosimetry tests, and cyclic triaxial tests were performed to determine the soil-water characteristics curves, pore size distribution, and p of compacted clay specimens with different moisture contents (w), suction (s), and number of FT cycles (NFT).For specimens at specific NFT, s and w, their moisture and temperature histories were imposed following three different paths, namely the WD-FT, FT-WD, and random paths that combine several interlaced WD and FT processes.The structural, soil-water, and dynamic characteristics obtained from specimens with different moisture and temperature histories were compared and analyzed.This paper provides experimental evidence for a better understanding of the hydro-mechanical characteristics of pavement subgrade soils under complex environmental conditions.

Testing Material
The soil investigated in this study was collected from Heilongjiang Province in north-eastern China, a seasonally frozen region.The collected soil samples were oven-dried, pulverized, and sieved (sieve opening=2 mm).The specific gravity Gs, liquid limit wL, plastic limit wP, plasticity index IP, optimum moisture content wopt, and maximum dry unit weight γdmax are 2.69, 45%, 23%, 22%, 22.76%, and 15.78 kN/m 3 respectively.The soil contains 10% sand, 73% silt, and 17% clay.Based on the "Test methods of soils for highway engineering JTG 3430-2020" [8], this soil is classified as low liquid limit clay.

Testing Methods
Two types of specimens were statically compacted at the soil's wopt and γdmax (i.e.disc specimens with a diameter of 61.8 mm and height 20 mm and cylindrical specimens with a diameter of 38 mm and height 76 mm).The cylindrical specimens were used to determine the εp and SWCC while the disc specimens were used for observing the soil structure.The compacted specimens were wrapped in two layers of plastic film for a minimum of 48 h to prevent mechanical damage IOP Publishing doi:10.1088/1755-1315/1330/1/0120643 and achieve moisture equalization.
Three types of moisture and temperature histories were imposed on the specimens following three different paths, namely the WD-FT, FT-WD, and random paths, as illustrated in figure 1. (i) Following the WD-FT paths, the as-compacted specimens were first wetted or dried from the wopt to 5 other moisture contents.The specimens after the WD processes were directly tested or subjected to 1, 3, or 10 FT cycles (NFT=1, 3, 10) and tested; (ii) In the FT-WD paths, the sequence of WD and FT processes was reversed.The as-compacted specimens were first subjected to 1, 3, or 10 FT cycles and then wetted or dried to obtain the same moisture content as adopted in the WD-FT paths; (iii) the random paths combine several interlaced WD and FT processes.Specimens were first subjected to 1 FT cycle at wopt and then went through different WD and FT processes until the designated w values at NFT=10 were achieved.The WD and FT paths (indicated in figure 1c as a sequence of numbers) are chosen to impose different WD and FT processes on each specimen.Specimens with the same w and NFT may have experienced two or three different M-T histories.
Consequently, the influences of M-T histories on the soil structure, SWCC, and p can be investigated through comparisons between the results of parallel specimens.The εp of the cylindrical specimens were determined through cyclic triaxial tests under one level of confining stress (c=13.8kPa) and three levels of deviator stresses (d=27.6,41.4, 68.9 kPa).
Each specimen was loaded 5000 times under three different stress conditions (that is three combinations of c and d).Following the cyclic triaxial tests, the specimens were sliced into smaller discs to measure the suction using the filter paper method [7,9].The moisture content-suction relationships obtained from the cylindrical specimens constitute the soil's SWCC.MIP tests were carried out on small clods sampled from disc specimens with different M-T histories.

Microstructural characteristics
Figure 2 illustrates the MIP results of specimens without FT history and at three moisture contents.Figure 3 illustrates the MIP results of specimens that have gone through FT-DW, DW-FT, and random paths at NFT=10 and w =20.76%.The cumulative mercury injection curves (CI curve) in the figures represent the relationship between the intruded mercury void ratio (eMIP, the ratio of cumulative mercury injection volume to the volume of soil particles) and pore diameter (d), while the pore size distribution curve (PSD curve) is the partial derivative curve of the CI curve (i.e. the relationship between -δeMIP/δlogd and logd).Compacted clay typically demonstrates an aggregated structure with identifiable aggregates and inter-aggregate pores [10].Following the method adopted by Han et al. (2022) [10] , the boundary between the structural pores and textural pores was chosen at d=10 μm where the slope of the CI curve changes significantly.
For specimens that have not undergone FT cycles (figure 2), the PSD curves at different moisture contents demonstrate unimodal characteristics, which show peaks in the range of the textural pores (d between 2 μm to 5 μm).Textural pores are sensitive to WD processes.They shrink during desaturation and expand during saturation.The peaks of the PSD curves change significantly with w.Meanwhile, the structural pores do not exhibit significant changes during the DW process, which is consistent with similar experimental observations in the literature [2].
Following 10 FT cycles, the PSD curves of specimens transformed from an unimodal form to a bimodal form (Figure 3), and a new peak appears in the range of structural pores.More structural pores and cracks appear owing to moisture and temperature effects.New peaks in the range of structural pores are generated, which are associated with the growth of ice crystals during FT cycles.The shapes of the CI and PSD curves of specimens with different M-T histories are similar.Therefore, it can be concluded that the influence of different moisture and temperature histories on the structure of compacted clay is consistent and similar.

Soil-water characteristics
Figure 4 illustrates the soil-water characteristic curves (SWCC) of clay.The symbols represent the measurements while the curves represent the fittings using the VG model [11] (equation ( 1)).
w  w sat [1 ( s a w where s is suction, wsat is saturated water content, aw is the parameter controlling the air entry value of the SWCC, and nw is the parameter controlling the shape of the SWCC. As the number of FT cycles increases, the SWCCs show two types of changes: (i) SWCCs move downwards, which means that the moisture content corresponding to the same suction decreases, and the suction corresponding to a given moisture content decreases; (ii) The slopes of SWCCs in the transition zones slightly increase.These changes are most noticeable after the first FT cycle.These phenomena indicate that the water retention capacity of soils decreases after the FT cycles, which is consistent with the observations in literature [4].Soil-water characteristics of soils are determined by their microstructures.In the transition zone, water is mainly stored in the textural pores [10].According to the MIP test results, after experiencing the M-T processes, structural pores in soils develop and textural pores shrink.Therefore, as textural pores decrease owing to M-T effects, the water retention capacity decreases, and the desaturation rate increases.Figure 4 illustrates that the suction of specimens at a given moisture content, albeit subjected to different M-T paths, exhibits relatively close.In all the experimental data, the difference in water content at the same suction and the difference in the suction for the same water content are within 10%.This observation suggests that following prolonged exposure to moisture and temperature effects, there is no significant difference in the SWCCs of specimens to different moisture and temperature paths.This finding aligns with the results of MIP experiments.

Accumulated plastic strain of compacted clay
The development of the accumulative plastic strains (εp) with the number of load cycles (N) for specimens undergoing FT-DW paths are illustrated in figure 5 (results of specimens undergoing DW-FT and random paths are similar).The εp accumulates mainly in the initial stage of cyclic loading.As N increases, the rate of increase in εp (i.e.slope of the εp-N relationship) gradually decreases and eventually stabilizes.Under the same condition, εp increases with increasing NFT, moisture content, and deviator stress, indicating that the mechanical properties of soils weaken during the FT processes of wetting, and repeated loading.
Werkmeister et al. [12] categorized the εp-N relationships into three types: plastic shakedown, plastic creep, and incremental collapse based on the difference between εp at 1330 (2024) 012064 IOP Publishing doi:10.1088/1755-1315/1330/1/0120646 the 5000th loading and εp at the 3000th loading (denoted as εpa) (equation ( 2)).The εpa reflects the slope of the εp-N relationship when the εp approaches stability.In this study εpa is considered as a key parameter for characterizing the permanent deformation characteristics of the compacted clay and studying its variation owing to temperature and moisture actions (figure 6).
  incremental collapse： pa  0.410 For specimens with different temperature and moisture histories, the εpa increases with the moisture content, deviator stress amplitude, and FT cycles.The relationship between εpa and w is nonlinear.When the water content is low (w≤wopt), the growth rate of εpa with w is relatively slow.When the water content is high (w>wopt), the εpa significantly increases with w.Meanwhile, the slope of the εpa-w curve increases with NFT, indicating that accumulative plastic strain becomes more sensitive to changes in moisture content after the FT cycles.pa Figure 7 illustrates relationships between the εpa and initial suction of the specimens (determined by the measured SWCC in figure 4).Compared with the results in figure 6, the εpa increases with decrease in suction, and the εp-N relationship gradually changes from plastic creep to incremental collapse.As the NFT increases, the εpa-logs relationship becomes steeper, indicating that εp becomes more sensitive to moisture.Instead of the nonlinear εpa-w relationship, the overall εpa-logs relationship exhibits a linear characteristic after temperature and moisture processes, especially when NFT≥3.Linear fittings of the εpa-lgs relationship at NFT=3 and 10 under different stress states show that its slope (λ=dεpa/dlogs) remains stable at low σd (27.6 kPa and 41.4 kPa), and significantly increases when σd increases to 68.9 kPa.There is apparently a cyclic load threshold (in this study, the threshold is between 41.4 kPa and 68.9 kPa).When the σd exceeds this threshold, the εpa becomes sensitive to moisture changes.
The results depicted in figures 6 and 7 lead to a conclusion that the variation in εpa with w, s, and σd in the samples subjected to different temperature and moisture paths exhibit certain differences during the FT cycle (NFT=1).However, when temperature and moisture effects reach equilibrium (NFT=10), the difference in the εpa in the samples along each path is relatively small.Therefore, it can be considered that under long-term environmental load, the variation in εpa is not closely related to temperature and moisture histories.

Summary and conclusion
In this study, the soil structure, soil-water characteristics, and accumulative plastic strain (p) were determined for compacted Heilongjiang clay that has experienced three different types of moisture and temperature histories.Experimental results and associated interpretation reveal the following conclusions: (i) Upom experiencing the temperature and moisture effects, structural pores of compacted clay develop, while textural pores shrink, resulting in a decrease in water retention capacity and an increase in desaturation rate.After long-term temperature and moisture effects reach equilibrium, specimens that undergo different temperature and moisture paths exhibit similar microstructure and SWCCs.
(ii) The p and its accumulative rate pa increase with NFT, w, and σd.Compared to the low moisture content state (w≤wopt), the p is more sensitive to moisture changes in the high moisture content state (w>wopt).After the freeze-thaw cycles, the sensitivity of p to moisture significantly increased.This is related to the changes in t h e soil-water characteristic of the compacted soil after the action of temperature and moisture, as well as the weakening of soil structure and connection strength between soil particles.
(iii) After experiencing the temperature and moisture influences, the pa-w relationship of the compacted soil exhibits nonlinear characteristics, while the εpa-logs relationship exhibits

Figure 2 .
Figure 2. (a) CI and (b) PSD curves of specimens at various moisture contents.

Figure 4 .
Figure 4. SWCC of specimens with different moisture-temperature paths.

Figure 5 .
Figure 5. Development of the εp during cyclic loadings.The boundaries between the different types of εp-N curves are marked in figure 6 based on equation (2) (where B refers to plastic creep and C refers to incremental collapse).Nearly all the εp-N curves of the tested specimens belong to the plastic creep or incremental collapse type.When the moisture content, deviator stress, or NFT increases, the εp-N relationship changes from plastic creep to incremental collapse.When moisture content and deviator stress are high, the εp-N relationship of compacted clay after FT cycles is more likely to become incremental collapse.

Figure 6 .
Figure 6.The variation of the εpa with w of specimens with different M-T paths.

Figure 7 .
Figure 7. Variation of εpa with the s of specimens with different M-T paths.