The test apparatus, as is displayed in Figure 1 in detail, was comprised of an electrokinetic cell and a D.C. power supply device. The electrokinetic cell, made of acrylics, holds a dimension of 440 mm in length, 330 mm in width, 140 mm in height and 10 mm in thickness. Similar to the experiment disposition designed by Chien et al. (2010), tubular stainless steel tubes were used as both electrodes (350 mm apart from each other) and the central tube (a tube at the midpoint between the anode and the cathode, 175 mm away from both electrodes), while the holes were drilled along the surface of the tube for injection (anode and central tube) and draining (cathode) during the electroosmotic process. The D.C. power supply device, which can provide an output voltage of up to 60 V and a current of 5 A, was connected to both the anode and cathode tube for electric supply. A number of drainage holes with a diameter of 3 mm were scattered at the bottom of the cell with a spacing of 10 mm from the cathode. The drainage process could be controlled through a drained tube at the end of the cell during the test.

Five voltage probes were installed on top of the cell to measure the voltage following an organized time interval. In the meantime, the vertical deformation resulting from soil consolidation is monitored by five dial gauges disposed on the top plater. Figure 1 illustrates the specific position of both voltage probes and dial gauges. The measuring cylinder and the multimeter were utilized to monitor of drained water volume and voltage, respectively. In addition, a camera was applied during the electroosmosis tests to investigate the physical behavior of soil-anode interface.

The soils used in this study were collected from the Jiangning District of Nanjing, China. The physical properties of the soil were assessed in accordance with the Chinese Standard GB/T 50,123-2019 (Standard for Soil Test Methods) (Chinese Standard GB/T 50123-2019, 2019) and were summarized in Table 1. Based on the Unified Soil Classification System (USCS) (ASTM International Standard D2487-17e1, 2017), the soil was classified as low plasticity clay (CL). Additionally, calcium chloride (CaCl₂) solution was employed as the injection material during the electroosmosis process.

A specified amount of air-dried soil was first mixed with distilled deionized water using a mechanical mixer to reach a water content of 1.5 times the liquid limit. This mixture was then stored in an airtight container with a sealed lid for 3 days to facilitate moisture equilibration. Subsequently, the prepared soil sample was layered into the electrokinetic cell in five separate layers, allowing for the placement of electrodes and a central tube. A saturated geomembrane was then positioned over the soil surface to create a horizontal flow condition.

A direct current of 30 V was applied to the soil with a voltage gradient of 50 V/m to undertake the electroosmotic process, with each test lasting for 25 h. A total of 140 mL of CaCl₂ solution with concentration of 2 mol/L was injected into the anode or the central tube, and water was discharged from the cathode during electroosmotic process. The voltage, current, surface settlement and drained water from the cathode were monitored during the test. The penetrometer resistance of the soil sample was assessed using a specially designed laboratory micro penetrometer, as illustrated in Figure 2. This micro penetrometer featured three dynamometers (I, II, and III) and three probes (A, B, and C). The actual penetration resistance can be obtained by multiplying the recorded reading by the corresponding calibration coefficient, with a measurement accuracy of ±5%. After the experiments, penetrometer resistance values were measured at various locations within the sample, as shown in Figure 2. Data readings were recorded electronically using a digital data logger. A total of four types of test were performed. The procedures for each type of tests are described below.

Four electroosmotic tests were denoted as S0, S1, S2 and S3 respectively. As presented in Figure 3, S0 refers to pure electroosmotic flow, S1 refers to electroosmotic flow with simultaneous injection through both the anode (Point A) and central tube (Point B), S2 refers to electroosmotic flow with anode movement (from Point A to B) only, and S3 refers to electroosmotic flow with anode movement (from Point A to B) coupled with injection of CaCl₂ solution (through Point A followed by Point B). Specifically, for test S1, 70 mL of CaCl₂ solution was firstly injected into the anode (Point A) and central tube (Point B) simultaneously immediately after powering on, and the electroosmotic process will continue for the whole treatment time of 25 h. For test S2, a pure electroosmotic process was first conducted for 12.5 h, after that the anode was pulled out and moved from Point A to Point B, followed by another 12.5 h of electroosmosis. While for test S3, 70 mL of CaCl₂ solution was firstly injected into the anode (Point A) immediately after powering on, then the electroosmotic process will last for 12.5 h. After that, the anode was pulled out and moved from Point A to Point B, and another 70 mL of CaCl₂ solution was injected into the anode (Point B), then the electroosmotic process will last for another 12.5 h. A detailed schematic diagram of these four tests are tabulated in Figure 3.

Figure 4 illustrates the variation of drained water (mL) during the electroosmotic process. The total volume of the drained water was 295, 609, 432, and 1030 mL for test S0, S1, S2 and S3, respectively. As shown in the figure, the drainage volume for the S3 test was the largest, approximately 3.5 times that of S0 test, 1.6 times that of the S1 test and 2.4 times that of the S2 test. It is worth noting that for S2 and S3 tests, a turning point in the drainage curve is observed around 12.5 h. In the first 12.5 h, the drainage volume was 170 mL for S2 and 496 mL for S3 test. Following this turning point, the drainage volume increased to 262 and 534 mL for S2 and S3 test, respectively.

Figure 5 shows the drainage rate (mL/h) vs. time (h) during the electroosmotic process. The initial drainage rate for the four tests was approximately 30 mL/h. During electroosmotic process, the drainage rate of S0 exhibited a continuous decrease until it reached a state of stability. For S1 test, the drainage rate initially increased rapidly and reached a peak drainage rate value of 70.6 mL/h at 2.25 h. This was followed by a slow decrease in the next 13–15 h, after which a stable stage was reached. In comparison, both S2 and S3 curves exhibited a subsequent peak around 12.5 h. For the S3 test, the drainage rate increased from 28.5 to 49.4 mL/h in the first 2.5 h, followed by a gradual decrease and a subsequent peak of 57.2 mL/h was reached around 12.5 h. Conversely, the S2 test showed a continuous decrease in drainage rate from 26.0 to 12.4 mL/h in the first 12.5 h and reached its maximum value of 42 mL/h when the anode was moved, followed by a downward trend thereafter. Yoshida (2000) and Burnotte et al., (2004) showed that with the continuous application of a direct current, the electrical contact resistance between the electrodes (mainly the anode) and the soil is considerably increased, leaving an effective voltage gradient too small for significant electroosmotic dewatering. Therefore, it can be concluded that the increase in drainage rate of S2 and S3 during the subsequent 12.5 h was attributed to the movement of the anode, which effectively doubled the voltage gradient, thereby enhancing the drainage capacity of the soil matrix. In contrast, under the same condition of anode movement, S3 exhibited better drainage performance than S2. This is mainly because the injection of CaCl₂ solution during the electroosmotic process in S3 will lead to an increase in electric conductivity and hydration of cation, resulting in a better drainage effect.

Figure 6 illustrate the variation of electric current (A) during the electroosmotic process. As noticed in Figure 6, the electric current in S0 test continuously decreased over time. Additionally, the electric current of S1 showed a rapid increase to more than 0.5 A due to the presence of salt solution, as well as the desorption and mobilization of ions in the soil matrix. Subsequently, the electric current decreased from 0.53 A to 0.06 A and finally a stable stage was reached. The decrease in electric current can be attributed to two reasons: first, a decline in the gradient of ionic concentration and a partial saturation of charge sites within the clay (Yukselen-Aksoy and Reddy, 2012), and second, anode corrosion and the formation of cracks near the anode area, which caused significant voltage loss and further reduced the electric current (Wu et al., 2015; Shi and Zhao, 2020; Shi et al., 2021b). For S2 and S3 tests, the change in electric current curve was similar to that in drainage rate curve, with a subsequent current peak value of 0.21 A for S2 and 0.42 A for S3 around 12.5 h. This observed increments in drainage efficiency and electric current around 12.5 h of S2 and S3 tests were may be due to fact that the anode being moved during the electroosmotic process. As mentioned, the movement of the anode, coupled with the injection of CaCl₂ solution can significantly decrease the power loss and double the voltage, thereby leading to a substantial increase in electric current. In addition, the peak value of electric current of S3, which is twice that of S2, is mainly due to the fact that injection of the CaCl₂ solutions.

A plot of electric resistance of the soil with regard to time is presented in Figures 7B–E. To provide a more comprehensive analysis of the electric resistance variations during the electroosmotic process, the changes in electric resistance for Part 1 (the soil adjacent to Point A), Part 2 (soil near the left side of the midline/Point B), Part 3 (the soil near the right side of the midline/Point B), and Part 4 (soil adjacent to Point C) were examined, as depicted in Figure 7A. Overall, the soil sample can be delineated into two main regions by the midline: the upper regions, which include Part 1 and Part 2, and the lower regions, which include Part 3 and Part 4.

As illustrated in Figure 7B, the electric resistance of S0 test rapidly increased over time, reaching a value of 1,200 Ω after 25 h in Part 1. In comparison, the electrical resistance of S1 was obviously lower than that of S0 in the same section. This is attributed to the injection of CaCl₂ solution, which markedly increased the total ion concentration within the soil matrix, resulting in an increase of current flow and a corresponding decrease in electric resistance. Comparing S1 and S3 tests in Figure 7B, it also can be observed that under the same treatment condition (electroosmotic process with injection of CaCl₂ solution), the electric resistance of both tests remained at a relatively low level (below 60 Ω) in the first period of 12.5 h in Part 1. Subsequently, S1 experienced a significant increase in electric resistance, surging from 55 Ω to more than 300 Ω, while the electroosmotic process of S3 was halted due to the movement of the anode. Observing the electric resistance curves in Figure 7C, it can be noted that the electric resistance of S2 increased progressively after 12.5 h. In contrast, under the same conditions of anode movement, the resistance of S3 initially decreased and then experienced a slight increased, while it consistently remained below 55 Ω. Therefore, it can be concluded that injecting of CaCl₂ solution coupled with moving the anode can significantly reduce the growth of electric resistance in Part 3. It is also worth noting that the electric resistance in the S2 and S3 tests was slightly higher than that in S0 and S1 within the regions of Parts 2 and 4, with the increase being generally less than 60 Ω, which is considered within an acceptable range for the tested soils. Clearly, the electric resistance was mainly affected in the regions of Parts 1 and 3 because the soil near the anode was more prone to water loss. In addition, water electrolysis and the impervious boundary led to an increase in gas pressure at the anode-soil interface, leading to significant volume shrinkage and cracks in both the Part 1 and Part 3 regions. These shrinkage cracks and gas accumulations substantially elevated the electric resistance, further reducing the electric current. Therefore, coupling anode movement with the injection of calcium chloride (CaCl₂) solution during electroosmotic process can effectively prevent the increase in electric resistance, which further enhance the electroosmotic treatment efficiency.

High power consumption limits the engineering applications of electroosmosis. To evaluate the power consumption of the three tests, the electrical power consumption during the electroosmotic process and the unit drainage energy consumption were calculated and analyzed.

The electrical power consumption, W , can be calculated using: W = ∫ V ⋅ I ⋅ t 1000 d t

Where V is the applied vlotage (V), I is the electric current as a fuction of time (A), and t is the processing time (h).

Figure 8 shows the power consumption (kW∙h) vs. time (h) during the electroosmotic process for the three tests. The power consumption after treatment is also presented in Table 2. The final power consumption was 0.05, 0.20, 0.09, and 0.21 kW h for tests S0, S1, S2 and S3, respectively. Analysis of the results from the S1 and S3 tests indicated that moving anode during the electroosmotic process had no significant impact on power consumption. Conversely, the power consumption following the injection of CaCl₂ was approximately twice that of the condition without any solution injection. This is mainly because the electroosmotic flow with injection will result in an increase in electric conductivity, thus the power required for the process increases proportionally (Mohamedelhassan and Shang, 2001). Although the power consumed in tests S1 and S3 was slightly greater than that in tests S0 and S2, but the increment is acceptable in the case of the tested soils.

Figure 9 illustrates the variation of unit drainage power consumption (mL/(W∙h)) with time during treatment. The unit drainage power consumption, η, was obtained by the volume of drained water (mL) by dividing power consumption (W∙h). As shown in the figure, the η of S0 fluctuated around 7 mL/(W∙h) over time, while the η of S1 initially increased rapidly, peaking at 2.25 h with a value of 6.1 mL/(W∙h). This was followed by a gradual decrease in the next 13–15 h and afterwards a stable stage. The changes in η of S2 and S3 in this study was quite different from those of S0 and S1. In the first 12.5 h, the η for S2 and S3 tests exhibited a similar trend, with minor fluctuations around 8 mL/(W∙h) over time. Subsequently, both S2 and S3 experienced a notable reduction in η, with values decreased from 8.0 mL/(W∙h) for S2 and 67 mL/(W∙h) for S3 to 3.4 mL/(W∙h) and 2.3 mL/(W∙h), respectively. In summary, the η values for S2 and S3 were consistently higher than that of S1 test throughout the entire electroosmotic process, indicating that S1 test requires more electrical power to discharge the same volume of water.

The final settlement measured at five locations, namely 0, 8.75, 17.5, 26.3 and 35 cm after treating vs. the distance from the anode for four different test procedures are presented in Figure 10. It is not surprising that all the settlements of five locations treated by test S0 keep the lowest, which was consistent with the previously observed lowest drained water among the four tests as shown in Figure 4. It can be readily seen that the settlement near Point A induced by test S1 was close to 6 mm, which is almost twice than that induced by S2 and S3, indicating that keeping the anode at Point A is benefit for the increase of final settlement near Point A. However, throughout the testing area, the settlement induced by S3 always keeps the largest except the observed Point A. To be specific, the final settlements measured at 8.75 and 26.3 cm away from Point A induced by S3 are 9 and 8.5 mm, 1.5 times than those induced by S1 and triple than those induced by S2. It can be concluded that moving the anode from Point A to B can certainly improve the settlement distribution throughout the whole treating area, especially for the area between Point A and B as well as Point B and C. It is interesting that the final settlement measured at Point B induced by S2 was lower than that induced by S3 and larger than that induced by S1, which means that the improvement on settlement brought by moving the anode from Point A to B is more drastic than that brought by electroosmotic flow with injection. It should also be admitted that the final settlements measured around cathode (Point C) induced by S1, S2 and S3 were close to each other and show not too much improvement compared with S0.

Penetration resistance can describe the strength of soils after electroosmotic treating. Figure 11 shows the contour plots of penetration resistance for the whole treating area after treatment for four tests. The contour plots were generated using Kriging method (Liang et al., 2018), which is a spatial interpolation estimator that is applied to find the best linear unbiased estimate at each location and is determined according to the linear combination of the known values of all sampled locations. The initial penetration resistance for natural samples ranged from 50 to 75 kPa for all the four tests. It is readily seen that Figure 11 can be divided into two categories, i.e., Figures 11A, B present the results for electroosmotic flow without anode moving, while Figures 11C, D present the results for electroosmotic flow with anode moving. It is obvious that the contour lines around Point B induced by S2 and S3 were denser compared with those induced by S1, indicating that moving anode from Point A to B could significantly improve the penetration resistance of soils around Point B. Furthermore, it is apparent that the soils, not only between the anode and the cathode, but also away from the alignment of the anode and the cathode, were significantly improved in S3. It should be noted that the penetration resistance of soils around Point A would be weakened by moving the anode from Point A to B. For example, the peak penetration resistance of soils around Point A treated by S1 was more than 1,400 kPa, while the peak values for S2 and S3 were about 500 kPa and 125 kPa, respectively. However, it has been proved that the area of penetration resistance greater than 100 kPa after treatment could be defined as a cementation area to describe the improvement of electroosmotic treatment, similar to the method used by Chien et al. (2010). It is expected that cementation between soil particles due to the chemical reaction between injected solutions should contribute to a large proportion of the cone resistance when it is greater than 100 kPa. Figure 12 plots the cementation area after treatment for four tests. It was calculated from Figure 12A and c that the cementation area for S0 and S2 were about 18.05% and 26.76% of the whole treating area, respectively. It is also evident that the cementation area in S1 and S3 were larger than that in S0 and S2, accounting for about 62.30% and 63.20% of the total area for S1 and S3, respectively. It is interesting that the cementation area treated by S1 and S3 were very close, the main difference lied in that the cementation area treated by S1 covers most of the area between Point B and C, whereas the cementation area treated by S3 covers most of the area between Point A and B. This is mainly because the injection of CaCl₂ solution into Point B was kept during the treatment time for S1, while for S3, CaCl₂ solution was injected to Point B only after the abode was moved from Point A to B. This observation indicates that electroosmotic process with anode movement coupled with injection of CaCl₂ solution could further improve the homogeneous distribution of the treatment area.