Investigation on Numerical Simulation of Chloride Transport in Unsaturated Concrete

College of Civil Engineering, Henan University of Technology, Zhengzhou 450001, China Key Laboratory of Performance Evolution and Control for Engineering Structures of Ministry of Education, Tongji University, Shanghai 200092, China School of Qilu Transportation, Shandong University, Jinan 250002, China Department of Architecture and Civil Engineering, Wenzhou Institute of Technology, Wenzhou 325035, China


Introduction
Concrete structures in coastal areas are often directly exposed to a high chloride environment during service life. Under the long-term impact of salt spray, the chloride content in concrete reaches the threshold level, which leads to the corrosion of steel bar in concrete structure, and the reduction of safety performance [1][2][3][4]. Chloride penetration also causes a decrease in the durability of concrete structures [5][6][7]. e main driving force of chloride transport in concrete is diffusion and capillary action [8,9]. In addition, factors such as temperature, RH, surface chloride content, water--cement ratio, and mineral admixtures affect the chloride transport process in concrete [10][11][12]. e Life-365 calculation program developed by the 365 Committee of the United States put forward the equation of the relationship between the chloride diffusion coefficient for ordinary Portland cement concrete and the water-cement ratio [13]. e study performed by Zhang et al. [14] confirmed that RH played an important role in the chloride diffusion coefficient; that is, the chloride diffusion coefficient increased with RH increasing. Zhu et al. [15] explored the effect of environmental RH on the permeability of chloride in unsaturated concrete. e results showed that when RH was greater than 75%, the content of chloride in concrete increased gradually, while the content of chloride on the surface decreased gradually over time. Shen et al. [16] further researched and found that a higher RH accelerated the migration of chloride in the concrete. Qi et al. [17] analyzed the variation of chloride diffusion coefficient with temperature using the Nernst-Einstein equation and then established the diffusion coefficient equation in the function of temperature. Matsumura et al. [18] found that the chloride diffusion coefficient became larger with higher temperature, and the logarithm of diffusion coefficient was linearly related to the reciprocal of temperature. Environmental RH influences the process of chloride transport and further impacts the transport rate and corrosion depth of chloride in concrete.
Liu et al. [19] reported that the time of chloride salt erosion and the erosion methods (dry-wet cycle, continuous erosion) affected the chloride diffusion coefficient. Li and Li [20] proposed a chloride transport model in cover concrete under drying-wetting cycles. Chen and Liu [21] found that the water migration rate in the drying stage was much less than the wetting stage. For this, Lin and Jiang [22] considered the water transfer speed in the process of moisture absorption and desorption of concrete and established the water transport equation under the drying-wetting cycle environment by adopting multiple water diffusion coefficients. On the basis of the water transport equation, Lin et al. further put forward the chloride transport equation under the drying-wetting cycle environment. Based on the diffusion of water transport in concrete and the permeation of water, Sun et al. [23] obtained a chloride transport model in unsaturated concrete and found that the chloride content in concrete was independent of the dry-wet ratio. Sun et al. [24] studied the effects of water-cement ratio, NaCl solution concentration, and exposure time on chloride diffusion coefficient, penetration depth, and content in concrete under dry-wet circulation.
From the above literature, it can be concluded that the moisture transport rate of concrete during moisture absorption and desorption processes is not consistent with RH. e fact is that chloride diffusion and capillary are very dependent on environmental temperature and humidity. However, the effects of concrete capillarity on chloride transport in different temperatures and RHs have rarely been investigated. us, it is necessary to establish a chloride transport model to express the influence of temperature and humidity in different moisture absorption and desorption environments so that the mechanism of unsaturated chloride transport could be better demonstrated. erefore, this study investigated the influence of RH and temperature on capillary pressure in water transport using concrete moisture absorption and desorption tests and then determined the parameters of the water transport model. e variation of chloride diffusion coefficient with temperatures and time was evaluated by the chloride penetration test. A chloride transport equation was established in function of capillary action and diffusion. Further, the effects of saturation on chloride transport were simulated by the numerical model.

Test Materials.
e concrete specimens were made of cement, mineral powder, fine aggregate, and coarse aggregate. e binder material comprised P.O. 42.5 cement and mineral powder with densities of 3060 kg/m 3 and 2890 kg/m 3 , respectively, in equal mass fractions. e chemical compositions of cement and mineral powder are shown in Table 1. Fine aggregate is well-graded medium sand with a practical density of 2623 kg/m 3 ( Table 2). Coarse aggregate is 5∼20 mm continuous graded gravel with a maximum particle size of 20 mm (Table 3). Its practical apparent density is 2710 kg/m 3 . e mix proportion of concrete in the chloride diffusion test is shown in Table 4. e W-2 mix proportion concrete mixture ratio of the moisture absorption and desorption equilibrium test is the same as that in Table 4. e concrete specimens in the dimension of 100 mm × 100 mm × 100 mm were prepared and cured at 20 ± 2°C and 95% RH for 28 days. At the age of 28 days, the compressive strength of cubic concrete specimens with the size of 150 mm was measured as shown in Table 4.

Moisture Absorption and Desorption Test.
A total of 40 specimens were cut horizontally into three pieces in the dimension of 100 mm × 100 mm × 30 mm. To avoid the test error, the middle piece was selected to test.
e test was carried out in accordance with ISO12571-2013 standard [25]. In the moisture absorption test, the specimens were first dried at 100°C for 24 hours. After the specimens were cooled to 20°C, they were weighed and recorded.
en, the specimens were dried for 12 h and weighed again until the mass loss was less than 0.5%. In the moisture desorption test, the dry specimens obtained according to the above test method were firstly soaked in water for 7 days to be saturated. Second, the surface of specimens was dried by towels and their weights were measured. e above two procedures were repeated until the weight loss was less than 0.5% in 24 hours to ensure a saturated specimen.
As seen in Figure 1(a), the middle part of the glass container is a porous ceramic partition. During the experiment, the testing specimen was placed on the ceramic partition. e temperature and RH under each test condition were monitored by a thermohygrometer in real time. Any change would be adjusted in time. e glass container was placed in a constant temperature and RH chamber (Figure 1(b)), with ±1°C and ±2.0% precision in temperature and RH, respectively. e designated RHs of 10%, 35%, 80%, and 95% in the glass container were controlled by LiCl, MgCl 2 , KBr, and KNO 3 saturated salt solutions, respectively. e 50% RH was directly controlled by the chamber. e test was carried out at the temperatures of 20°C, 35°C, and 50°C.
Hereafter, the specimens for moisture absorption and desorption test were put into glass containers which were moved to a constant temperature/humidity chamber. Isothermal absorption devices are shown in Figure 1(b). en, the mass of each specimen was measured periodically until the change in mass was less than 0.1%. Due to high moisture absorption (desorption) in the early stage, the corresponding time interval of measurement was small while it was large in the late stage. ere were 12 weight measurements for the absorption test and 13 weight measurements for the desorption test. e measurement results showed that, in the moisture absorption environment of different temperatures, the concrete specimens need 325 h and 100 h to reach equilibrium for RH >50% and RH <50%, respectively. In the moisture desorption environment of different temperatures, the equilibrium time was 150 h and 200 h for RH >50% and RH <50%, respectively.
For porous cement-based materials, moisture saturation can be used as a function of moisture content changes in mass [26]. e equilibrium saturation of concrete in different environments can be described as follows [27]: where S is the mass of specimen underwater (kg), m w is the mass of specimen underwater (kg), m d is the mass of specimen in a dry state (kg), and m ws is the mass of specimen under saturation state (kg).

Chloride Transport Test.
To better reflect the impact of environmental parameters on the chloride penetration in concrete, five surfaces of each specimen were smeared with epoxy resin to only allow chloride to penetrate from the other side. en, the specimens were placed in a salt spray box to simulate the chloride penetration environment by controlling the salt spray concentration and temperature. In the salt spray penetration test, the temperature for the specimens of W-1 to W-4 was 35°C. As a reference, temperatures of 20°C and 50°C were also used for W-2 specimens. e saltwater concentration in salt spray box was 50 g/L, and the test period was 60 days. e average of 4 specimens for each mix was reported.
e powder for chloride content test was drilled from the specimens every 20 days. After dried and weighted, the powder sample was put in deionized water and stirred evenly. e chloride content was measured by a rapid determination instrument.

e Equilibrium Relationship between RH and Saturation.
In the moisture absorption and desorption tests, the mass change rate of concrete is the equilibrium saturation under corresponding environment conditions when it reaches the equilibrium state. e equilibrium saturation of moisture absorption and desorption tests at 20°C, 35°C, and 50°C were calculated by equation (1). e equilibrium curves of RH and saturation in moisture absorption and desorption tests at different temperatures are shown in Figure 2.
In Figure 2(a), it can be seen that the trend of the equilibrium saturation curve is similar under three temperatures for absorption. Specifically, for a given temperature, the equilibrium saturation increases with the increment of RH. Moreover, regardless of RH, the equilibrium saturation of concrete gradually decreases as the temperature increases. e relationship between RH and maximum pore size can be determined by the Kelvin equation [28]. e maximum pore size that can be saturated is about 4 nm when RH is 50%. Generally, the boundary between gel pore and capillary pore is 10 nm (i.e., gel pores <10 nm, capillary pores >10 nm). erefore, when RH is less than 50%, only a few gel pores are saturated, while RH >50%, the capillary pores are gradually saturated [29]. It indicates that RH is dominant for equilibrium saturation of concrete when it is higher than 50%. When RH is below 50%, the temperature impact is not significant on saturation. When RH >50%, the saturation increases most at 20°C, and least at 50°C. It also shows that the effect of temperature on moisture absorption became considerable.
As seen in Figure 2(b), the trends of desorption are similar. For a given temperature, saturation decreases with RH decreasing. At 50°C, the trend is almost linear. Despite the same RH, the equilibrium saturation gradually decreases with the increase in temperature. It is found that the variation of saturation is limited when RH >80%. However, when RH <80%, saturation greatly decreases. It indicates that 80% RH is an inflection point for saturation in desorption.
It can be seen from Figure 2 that, under the same RH, the desorption curve with higher saturation has an obvious hysteresis compared with the absorption process. e main reason for the hysteresis is the "ink bottle effect," which reflects the effects of micropores (bottleneck) connected with the macropores (bottle body) on the flow of water in and out [30]. It is apparent that RH required for the macropores to reach saturation is higher than that for the micropores. For the initially saturated pores, when RH is gradually reduced, the liquid water in the large pores has to transfer through the small pores and could not form a gasliquid interface by itself. Hence, only when RH drops below the saturated RH in micropores, the water in the macropores could be discharged.
In the processes of moisture absorption and desorption, saturation is an important parameter to ascertain the water content of concrete. To represent the equilibrium relationship between RH and saturation in concrete, an absorption model of porous hygroscopic material between water absorption capacity and RH was adopted in [31]: where W is the equilibrium moisture content; h is RH; T is temperature (K); and A, B, and C are coefficients. e saturation of equilibrium moisture content was used instead of the equilibrium moisture content in equation (2), and equation (2) was turned into an inverse function; the relationship model between RH and saturation is acquired as where S is the saturation. According to Figure 2, parameters A, B, and C in equation (3) can be determined by nonlinear regression. e models of RH and equilibrium saturation in moisture absorption and desorption processes are shown as follows: For unsaturated concrete, capillary pressure is the main driving force for moisture transfer. Consequently, the relationship between capillary pressure and saturation is the key to investigating moisture transport [32,33]. Such relationship proposed in the literature is as follows [26]: where M is the molar mass of water (kg/mol), R is the gas constant (8.3144 J/mol/K), p c is capillary pressure (Pa), ρ l is the density of liquid water (kg/m 3 ). Substituting equation (6) with equations (4) and (5), the relationship between capillary pressure and saturation parameter during moisture absorption and desorption can be described as follows: Compared with saturated concrete, the pores in unsaturated concrete are not completely transporting fluid, and the permeability coefficient of water in unsaturated concrete is related to the degree of pore saturation. us, the relative permeability of moisture in the process of moisture absorption and desorption is relevant to the saturation of concrete. According to [34], relationships between capillary pressure and saturation as well as relative permeability and saturation can be established as where m can be determined by regression of capillary pressure and saturation curve. e relationship between capillary pressure and saturation at different temperatures in moisture absorption and desorption can be given by equation (9). e m values are shown in Table 5.
Afterward, the relationship between the relative permeability and saturation in the moisture absorption and desorption is established as follows: e m values under different temperatures in moisture absorption and desorption were calculated using equation (9). However, when equation (9) was used to ascertain A a values according to Table 5, the fitting correlation coefficient was only 0.56. erefore, it was not appropriate to use m to characterize the relationship between capillary pressure and saturation in equation (9). In absorption and desorption, equations (7) and (8) can be used to describe the relationship between capillary pressure and saturation.

Effect of Temperature on Chloride Content.
e specimen was subjected to a salt spray penetration test. e chloride content at different depths of concrete at 20 days, 40 days, and 60 days under 20°C, 35°C, and 50°C is shown in Figure 3. Figure 3 shows that the chloride content gradually decreases as the depth increases. e chloride content at different depths and the penetration depth of chloride increases over time. Regardless of time, the chloride content at a given depth in concrete shows the same trend; that is, the chloride content increases with the increase of  Advances in Materials Science and Engineering 5 temperature. It reveals that the increased temperature accelerated chloride penetration and enlarged the penetrated amount of chloride.

Effect of Temperature on Chloride Diffusion Coefficient.
Diffusion is the main driving force for chloride to penetrate saturated concrete. e stable diffusion of chloride can be described by the Fick second law [35]: where D Cl is the chloride diffusion coefficient (m 2 /s), C Cl is the chloride content (% of the mass of concrete), t is the time (s), and x is the depth from the concrete surface (m).
Assuming that the chloride content on the surface of the concrete structure is constant, D Cl is constant, and the initial chloride content at the exposed surface is 0. e following equation can be derived: where C S is the surface chloride content (% of the mass of concrete) and C 0 is the initial chloride content, which is assumed to be 0. e test results in Figure 3 were fitted by equation (13) to achieve the chloride diffusion coefficient at different temperatures and penetration times. e chloride diffusion coefficient with different temperatures and penetration time is shown in Figure 4. Figure 4 shows that the chloride diffusion coefficient gradually increases as the temperature increases, which is consistent with Matsumura's finding [18].
In Figure 4, the chloride diffusion coefficient at 40 d is significantly smaller than that at 20 d, and it is slightly increased compared with that at 60 d. is demonstrates that with temperature rising, its effect on the chloride diffusion coefficient becomes smaller. But in the temperature range from 20°C to 50°C, the chloride diffusion coefficient is increased by 37.8%. is is because the higher temperature tends to decrease the resistance of chloride penetration in concrete, which increases the rate of chloride penetration [36]. However, the chloride diffusion coefficient approach to a certain value gradually over time. ere are two models that describe the relationship between chloride diffusion coefficient and temperature. e one reported by Berke is called the Nernst-Einstein equation and the other is proposed in the American Life-365 standard design program. After the dimensionless processing and fitting comparison on the results in Figure 4, it is found that the model proposed in the American Life-365 standard design program is more suitable.
e proposed diffusion coefficient model is as follows [37]: where  Figure 5 shows the results of the salt spray test. e chloride diffusion coefficient decreases with time before 40 d. After 40 d, the chloride diffusion coefficient of W-1 and W-2 also decreases, while it increases for W-3 and W-4 with a larger water-cement ratio. e reason may be related to the discrete specimen itself. Overall, the chloride diffusion coefficient decreases over time.

Effect of Time on Chloride Diffusion Coefficient.
Based on a large number of experimental data, Mangat and Molloy [38] acquired a relationship of the chloride diffusion coefficient in concrete with time as follows: where t 1 is reference time, that is, exposure time of the first sample (s), t is exposure time greater than the reference time (s), D 1 is the chloride diffusion coefficient (m 2 /s) at t 1 , and α is the empirical coefficient. Figure 5 shows the chloride diffusion coefficient over time. Assuming that t 1 is 20 d, the relative ratio of the chloride diffusion coefficient is averaged under three temperatures for 20 d, 40 d, and 60 d. Equation (15) was fitted to determine the empirical coefficient α of 0.65.

Moisture Transport Model.
e main driving force for moisture transport in unsaturated concrete is capillary pressure. In [39], the relationship between the moisture transport velocity and capillary pressure is expressed by the Darcy-Buckingham equation as follows: where ] is the moisture transfer rate (m/s), k 1 is the effective transport permeability of water (m 2 ), which is related to the content of water in concrete and the microscopic geometric parameters of the pore, and µ is the dynamic viscosity coefficient of water (Pa·s). e mass of water flowing into the concrete is equal to the mass flowing out. Regardless of the moisture from liquid to vapor in the pores, the water transport equation can be expressed as [40,41] ] � − k μ k rl zp c zS grad(S).
e effective permeability of moisture transport can be expressed as

Advances in Materials Science and Engineering
where k is the intrinsic permeability of concrete (m 2 ) and k rl is the relative permeability, which is a parameter associated with the saturation. From equations (17) and (18), the moisture transport model in concrete can be described: where ϕ is the porosity of concrete and D (S) is the transmission coefficient of liquid water. e relationship between the dynamic viscosity coefficient of water µ and temperature T can be expressed as follows [42]:

(21)
Moisture transport consists of absorption and desorption. From equations (19) and (20), two basic parameters can be found: p c and S. To reflect the real moisture absorption and desorption transport process, the derivatives describing the relationship of capillary pressure and saturation in the process of water absorption and desorption can be obtained according to equations (7) and (8), respectively. Equation (21) was substituted into equation (20) to obtain the moisture transfer coefficient in moisture absorption and desorption, that is, equations (22) and (23)

Chloride Transport
Model. e chloride transport in concrete is essentially a coupling interaction of capillary chloride migration in water and chloride diffusion against concentration difference [22,43]. e chloride transport flux caused by capillary action is where J w,c is the chloride transport flux caused by capillary action (kg/(m 2 ·s)) and C w is the concentration of chloride in pore solution (kg/m 3 ). Meanwhile, the relationship between chloride concentration in the pore solution of concrete and free chloride content (percentage of concrete quality) is as follows: By substituting equations (17) and (25) into equation (24), the chloride transport flux caused by capillary action can be expressed as where C f is the chloride content in concrete (%) and ρ is the density of concrete (kg/m 3 ). e chloride diffusion transport flux caused by chloride content gradient in concrete can be expressed as where J d,c is the chloride transport flux caused by diffusion.
In concrete pore solution, ions can only be transported by diffusion behavior, which is closely relevant to concrete saturation. erefore, the chloride diffusion coefficient is largely relevant to the saturation of concrete. Climent et al. found a linear relationship between chloride diffusion coefficient and saturation in unsaturated concrete [44]. Guimarães et al. [45] also put forward the polynomial model, but when the concrete saturation is less than 50%, the reliability of the chloride diffusion coefficient calculated by this model will be reduced. e concrete saturation range established in the water transport model in this paper is large, so the linear model proposed by Climent is used: where D Cl (S) is the chloride diffusion coefficient in unsaturated concrete. Combining equations (14), (15), and (28), the equation of chloride diffusion coefficient in unsaturated concrete can be established: When chloride penetrates the interior of concrete, some ions will be chemically bound by the hydration products in cement to produce new hydrates, and some will be physically adsorbed by the pore surface of interior concrete. e binding capacity of chloride is mainly affected by the content of C 3 A in concrete, auxiliary cementitious materials, and PH value of pore solution. erefore, the total chloride content in concrete is the sum of free chloride content and bound chloride content, which is expressed as follows: 8 Advances in Materials Science and Engineering where C b is the bound chloride content in concrete (%). For chloride transport in concrete, it is necessary to quantitatively express the relationship between the free chloride content and the bound chloride content. Tuutti [46] found that linear binding could well describe such a relationship as follows: where β is a constant based on the cementing material and β is 0.15 according to a number of practical engineering surveys in [26]. e chloride transport in concrete is governed by chloride diffusion and moisture capillary and caused by chloride concentration gradient, water flow, and binding interaction. us, it can be expressed as

Heat Balance Equation.
e heat balance equation correlated to the saturation can be established based on Fourier's law [47] as where c is the specific heat of the material (J/kg·K), λ S is the thermal conductivity at different saturations (W/m·K), and Q is the heat flux (J/m 2 ·s). e negative sign indicates that the direction of heat flow is opposite to the direction of the temperature gradient. In this paper, the transient plane source (TPS) technique was used to determine the thermal conductivity of concrete for a given mix proportion at the same saturation values of 0, 0.3, 0.5, 0.7, 0.9, and 1.0 in the moisture absorption and desorption. e relationship between the thermal conductivity and saturation is where λ a is the thermal conductivity under dry condition (W/m·K).

Model Parameters and Mesh
Generation. e concrete is considered as a homogeneous and isotropic continuous medium. e solid skeleton of concrete is assumed as a rigid and invariant material which does not react with liquid phase and gas phase. Based on this assumption, the onedimensional chloride transport in concrete was explored.
According to equations (22), (23), (29), (32), and (33), a numerical transmission model was established by COMSOL ® software. e physical parameters of heat transport are shown in Table 6. e physical parameters of the moisture transport model and the chloride transport model are shown in Table 7.
e size of the moisture transport model is the same as that of the experimental specimen used in the test, namely, 30 mm × 100 mm × 100 mm. A total of 9520 tetrahedral elements were used to develop the finite element model with DOFs (degree of freedom) of 28,186 and convergence criteria of 10 −5 , as shown in Figure 6(a). Water can only be transmitted through the left boundary with the remaining surfaces closed, and there is no water transmission flux. e chloride transport model is a two-dimensional geometric model in size of 100 mm × 100 mm. e quadrilateral element is selected for the mesh division of this model. To make the solution domain of model denser, the quadrilateral element is divided by the detailed element size. A total of 578 domain elements and 48 boundary elements were used to develop the finite element model with DOFs of 28,186 and convergence criteria of 10 -5 , as shown in Figure 6(b). Chloride can only be transmitted through the left boundary, the remaining surfaces are closed, and there is no ion transmission flux.

Moisture Transport Model Validation.
e moisture absorption under three temperatures (20°C, 35°C, and 50°C) and two RH (80% and 90%) conditions was simulated based on the developed model and compared with the corresponding experimental results. e initial saturation and temperature were 0.08 and 283.15k, respectively. e simulation time was the absorption equilibrium time for the three temperatures, which was approximately 350 h. e moisture absorption simulated results and experimental results are shown in Figure 7. e saturation of concrete increases over time, and its growth rate is similar for different temperatures and RHs, which indicates that the simulated results are in good agreement with the experimental results. However, the experimental values are slightly higher than the simulated values. is is due to the existence of water condensation inside the concrete during the test. erefore, the moisture absorption in the simulated results is slightly lower than that in the experiment. e moisture desorption was simulated under temperatures of 20°C, 35°C, and 50°C and RHs of 10% and 35% and compared with the experimental results for validation. e initial saturation of 1 and temperature of 283.15k were used. e boundary saturation of concrete was equilibrium saturation with a simulation time of 250 h. Figure 8 provides the simulated and experimental results for desorption, which shows good coherence. For 10% RH, the difference between the simulated and the experimental equilibrium saturation increases with temperature increasing. When the concrete is in low saturation, the simulated moisture desorption rate is slightly lower than that of the experiment. It shows that water evaporation has little effect Advances in Materials Science and Engineering on the desorption rate of low saturation concrete, and its effect on equilibrium saturation increases with the increase of temperature. erefore, the simulated model is accurate.

Chloride Transport Model Validation in Saturated
Concrete.
e simulated and experimental results of chloride transport in saturated concrete of 0.46 water-cement     e size of the concrete specimen was 100 mm × 100 mm × 100 mm. e time of chloride penetration test was 60 days. e temperatures of 20°C, 35°C, and 50°C were used with a surface chloride content of 0.105%, 0.123%, and 0.233% as a boundary condition, respectively. e initial temperature was 293.15 K (0°C), and saturation was 1.0. e chloride penetration depth and chloride content distribution under three temperatures at 60 d were simulated.
It can be seen from Figure 9 that the simulated results are basically the same as the experimental results. e simulated chloride penetration depth is greater than that of the experiment. is could be explained as follows: (1) Bao et al.'s [48] study shows that the higher the relative humidity of the specimen-exposed environment leads to the greater water content within the specimen. e chloride diffusion was tested in a closed salt spray corrosion environment with a high RH, resulting in greater internal water content of the concrete specimens. At the same time, the condensation effect of water vapor occurred on the surface of concrete, forming a "water film" to dilute the chloride concentration, resulting in a reduced chloride concentration. us, the simulated chloride penetration rate and depth are greater than the experimental values. (2) In the experiment, the chloride content on the concrete surface gradually increased over time, and the results are consistent with the findings in [49], while in simulation, it is larger than the experimental value at 60 d. erefore, the simulated chloride content is greater than the experimental results.

Chloride Transport Model Validation in Unsaturated
Concrete. Concrete specimens with initial saturation of 0.7 were made with the same proportion as W-2. After curing, the concrete was cut horizontally into two 100 mm × 100 mm × 50 mm specimens for the chloride penetration test of unsaturated concrete. When the temperature and humidity became equilibrium, 20 g NaCl solid was placed on the top of the specimen. e specimens were wrapped and sealed with plastic film and were then put into a chamber (temperature � 25°C, RH � 50%). e chloride content in concrete was measured by a CLU-S type chloride content tester at 20 and 80 days.     Next, the simulated and experimental results of 0.7 concrete specimens were compared to validate the chloride transport model for unsaturated concrete. e boundary condition (x � 0) of chloride content is the experimental value of surface chloride content of 20 d and 80 d, which are 0.0344% and 0.0392%, respectively. Furthermore, the initial saturation of 0.7, the initial temperature of 293.15 K (20°C), and the chloride content of 0 were used for modeling. e comparison result is shown in Figure 10.
According to Figure 10, the trends of simulation coincide with that of the experiment, which effectively proves the reliability of the chloride transport model for unsaturated concrete. e chloride penetration depth was 12 mm and 20 mm at 20 d and 80 d, respectively. e experimental results were slightly smaller than the simulated results.

Capillary Effect on Chloride Transport in Moisture
Absorption.
e chloride transport in concrete with different saturation under moisture absorption was simulated to evaluate the effects of diffusion and capillary action on chloride transport. e concrete boundary conditions (x � 0) were set as follows: the temperature of the left terminal face of concrete � 308.15 K (35°C); the chloride content � 0.29%; and the saturation � 0.6, 0.8, and 1. e initial condition is as follows: temperature � 293.15 K (20°C), chloride content � 0, and saturation � 0.4. e simulation was calculated with the step length of 1 d and the total duration of 365 d. e distribution of chloride content at 50 mm depth under diffusion is shown in Figure 11, where PDS is pure diffusion saturation and CMAS is coupled moisture absorption saturation. e chloride content in the boundary conditions of the numerical simulation was based on the experiments, but it is not exactly the same as experiment results. It should be noted that the boundary chloride content (0.2%) at 293.15 K (20°C) was artificially set, which means that the influence of boundary chloride content on chloride transmission was studied. However, the boundary chloride content of 308.15 K (35°C) was calculated according to equation (35). Equation (35) ignores the influence of time on the surface chloride content as seen in Figure 3, which is fitted considering the ratio of temperature parameter (T/T 1 ) and the relative value of the surface chloride content (C T /C 1 ) with R 2 of 0.929: where T 1 is 293.15 K (20°C), C 1 is the surface chloride content corresponding to T � 293.15 K (20°C), and C T is the surface chloride content corresponding to the temperature T.
It can be seen from Figure 11 that, in moisture absorption, the chloride content is higher in the concrete with greater boundary saturation. However, the difference between chloride content under coupling action and diffusion gradually increases because the chloride transport under coupling action was the result of both diffusion effect and capillary action. erefore, the chloride transport speed under coupling action is greater than that under pure diffusion. is also indicates that the influence of capillarity action on chloride transport gradually increases with the increment of saturation gradient. When the saturation is 0.6, the effect of capillary action on chloride transport is minimal. At the same time, the difference between the chloride content under the coupling action and pure diffusion increases first and then decreases with the depth increasing. Regardless of saturation, the effect of capillary action on chloride transport increases firstly and then decreases with the increase of depth.

Capillary Effect on Chloride Transport of Saturation in Moisture Desorption.
e chloride transport in concrete with different saturation under moisture desorption was simulated. Comparing the diffusion and coupling transport (diffusion and capillary action) of chloride in concrete, the influence of coupling transport in the moisture desorption environment was assessed. Furthermore, the effect of capillary action on chloride transport was also investigated. e concrete boundary conditions (x � 0) were set as follows: the temperature of the left terminal face of concrete � 308.15k (35°C), the chloride content � 0.29%, and the saturation � 0.4. e initial condition was as follows: temperature � 293.15 K (20°C); chloride content � 0; and saturation � 0.6, 0.75, 0.9. e simulation was calculated with the step length of 1 d and the total duration of 365 d. e distribution of chloride content at 50 mm in the diffusion and hygrothermal coupling of concrete is shown in Figure 12, where PDS is pure diffusion saturation, CMDS is coupled moisture desorption saturation.
It can be found in Figure 12 that the content of chloride for pure diffusion increased slightly with the increase of saturation while the chloride content decreased gradually with the increase of saturation for coupling action.
Meanwhile, with saturation gradient increasing, the difference of chloride content between coupling action and diffusion is increased. e effect of capillary action on chloride transport is gradually enhanced with saturation gradient increasing. When the saturation is 0.9, the difference in chloride distribution between diffusion and coupling action is the largest, and then the effect of capillary action on chloride transport is the strongest. Meanwhile, for a given saturation gradient, the effect of capillary action on chloride transport increases first and then decreases with the increment of depth. e effect of internal saturation on chloride coupling migration and diffusion transmission is compared in Figures 13-15 at 256 days. It is found that, with the internal saturation increasing, the chloride content at the same depth under the coupling transmission increased while the chloride content at the same depth of pure diffusion transmission is not significant. When the internal saturation was the same, the chloride content of the same depth was larger in pure diffusion transmission. e reason for this phenomenon is that the effect of the concentration difference under diffusion transmission on the chloride content of the same depth became gradually stabilized, while the effect of capillary action on chloride transmission at the same depth was increasing.

Conclusions
In this paper, the diffusion and capillary flux of chloride in unsaturated concrete at different temperatures was experimentally investigated, and the relationship between capillary pressure and concrete saturation at different temperatures was achieved. Further, a transport model of temperaturehumidity-chloride was established. e influence of capillary pressure on chloride transport in moisture absorption and desorption was evaluated. From the experimental and simulated results, the following conclusions are drawn: (1) e inflection points of equilibrium saturation in isothermal moisture absorption and desorption for RH are 50% and 80%, respectively. In the moisture absorption (desorption), when RH is lower than (higher than) the inflection point, the temperature has little effect on the equilibrium saturation. But when RH is higher than (lower than) the inflection point, the saturation increases with temperature decreasing. (2) e increment of temperature cause the increases in chloride diffusion coefficient, and the increasing rate of chloride diffusion coefficient gradually reduces over time and finally stabilizes. (3) e simulated model is in good agreement with experimental results. However, it should be pointed out that the chloride transport model of unsaturated concrete has some defects and limitations in the lack of time-varying effect of chloride concentration. (4) e chloride transport velocity of the coupling environment is higher than that of pure diffusion in a moisture absorption environment with more chloride content found at the same depth. However, the opposite result is presented in the desorption process. Moreover, the effect of capillary action on chloride transport increased with saturation increasing [50,51].

Data Availability
e data used to support the findings of this study are included within the article.

Conflicts of Interest
e authors declare that there are no conflicts of interest regarding the publication of this paper.