Remediation of Benzene and 1,2-Dichloroethylene in Groundwater by Funnel and Gate Permeable Reactive Barrier (FGPRB): A Case Study

: Funnel and gate permeable reactive barrier (FGPRB) is an effective method to treat ground-water pollution. In order to clarify the impact of FGPRB on groundwater dynamic conditions, this study takes a site pilot test as the research object and establishes an FGPRB downstream of a petro-chemical industry. The results show that the concentrations of 1,2-dichloroethylene and benzene in the downstream groundwater, after setting FGPRB, are lower than the detection limit. The numerical simulation results show that after setting FGPRB, both point source and area source pollution can achieve a good delay effect, extending from about 27 d to about 65 d of response time, but changing the thickness and permeability coefﬁcient has no obvious effect on the delay effect. The tracer test shows the average permeability coefﬁcient of the medium from the injection well to the monitoring well after the construction of FGPRB decreases from 77.0 m/d to 31.2 m/d after the construction of FGPRB. The average seepage velocity from the injection well to the monitoring well decreased from 0.19 m/d to 0.078 m/d after the construction of FGPRB. At the same time, when the FGPRB is not built, the maximum concentration time from the injection well to the monitoring well is about 10 d. After the FGPRB is constructed, the maximum concentration time of the tracer received by the monitoring well is about 27 days. These results conﬁrm that the establishment of FGPRB will change the hydrodynamic conditions of groundwater and delay the response time of pollutants in the monitoring well.


Introduction
Petroleum refining usually refers to the distillation and separation of crude oil to obtain various products used in our daily life; it plays a significant role in the growth of the national economy [1].However, accidents in the production process usually lead to the pollution of soil and groundwater [2,3].Here, there are many kinds of chemical products involved in petroleum refineries, including total petroleum hydrocarbons (TPH), halogenated hydrocarbons, polycyclic aromatic hydrocarbons (PAHs), benzene series (BTEX), etc. [4].Most of the products are in-process enterprises, so once pollution occurs, it is very difficult to control.In the list of carcinogens published by the International Agency for Research on Cancer of the World Health Organization, petroleum refining is in the list of class 2A carcinogens [5].Therefore, the treatment of petroleum pollutants is necessary to human health.
Unlike water treatment, groundwater pollution remediation is very difficult.Traditional technologies, such as pump and treat (P&T), groundwater aeration gas, phase extraction, biological stimulation and in-situ chemical oxidation, are effective in remediation of groundwater pollution [6], but they have both advantages and disadvantages.For example, P&T has a remediation effect on groundwater pollution with a large pollution range and deep burial, but there are often tailing and rebounding phenomena, and the operation cost is high [7].In-situ aeration technology has the advantages of convenient equipment and instrument installation at low cost while at the same time promoting the degradation of pollutants by indigenous microorganisms.The disadvantage is that the targeted scope of application is limited [8].Biostimulation technology has a wide remediate range and low cost, but it has high requirements for permeability.When the permeability is poor, microorganisms will not be able to obtain sufficient energy materials and weaken their degradation performance of pollutants [9].In-situ chemical oxidation technology has a wide application range and high remediation efficiency but may introduce secondary pollution [10].Compared with the above technologies, permeable reactive barrier (PRB) is an important in-situ remediation technology for polluted groundwater, which has the advantages of no external power, no ground space and low operation cost, and has been applied for decades [11,12].The main influencing factors of whether PRB can achieve the expected remediation effect include [13,14]: (1) accurate pollution investigation; (2) effective repair materials and (3) reasonable response wall-size design.According to the 2002 USEPA report, most research on PRBs has focused on the use of zero-valent metals, especially metallic iron or ZVI granules, as a reactive substrate [12].In addition to ZVI, microorganisms, activated carbon, peat, montmorillonite, lime, sawdust and so on are also used as filling materials [15].The PRB device structure can be divided into continuous wall-type and funnel-gate-type [16].Although the funnel-gate PRB interferes with the flow field of groundwater to a certain extent, it can prevent the contaminated exudate from entering the downstream unpolluted area, which is more favored by environmental regulators.
However, most research focuses on the types of fillers in the PRB medium and the structural design of the wall, while little attention is paid to the changes in groundwater velocity and solute transport velocity in the long-term operation process [17].For example, Sarah et al. [18] proposed the need to evaluate the impact of PRB on groundwater flow and geochemistry after the completion of PRB in the feasibility study on the treatment of acid sulfate and nitrate-polluted groundwater, but did not conduct a detailed study in the field.Scott et al. [19] analyzed whether groundwater flow generated by the installation of PRB is significant to improve the design of PRB.However, their three-dimensional numerical simulation research only focused on how to optimize the capture of groundwater after the installation of PRB, and there is still a lack of research on the diffusion rate of pollutants after the installation of PRB.Therefore, in this study, an FGPRB was established downstream of a petrochemical site in Jilin Province, China.The objectives of this study were to: (1) evaluate the degradation effect of FGPRB on benzene and 1,2-dichloroethylene in groundwater; (2) simulate the change of pollutant response time of downstream monitoring wells after the establishment of FGPRB in different scenarios; (3) monitor and calculate the groundwater flow rate and pollutant diffusion rate during the operation of FGPRB.

Site Description
The study area is located in the Jilin Branch of China Kunlun Engineering Co., Ltd., No. 3-19 Xuzhou Road, Longtan District, Jilin City, Jilin Province, China.As shown in Figure 1, the study area is about 137 × 117 m rectangular, chemical plant area in the north and southeast, residential area in the west and highway in the southwest.The stratum structure in the study area is shown in Table 1, which mainly includes miscellaneous fill, silty clay, fine sand, gravel and granite.The groundwater flows from north to south, the hydraulic gradient is about 2.5‰, and the permeability coefficient of the aquifer (gravel layer) is about 77 m/d.The main source of contamination on the site is the chemical plant; there are many kinds of pollutants, among which the detection rates of 1,2-dichloroethylene and benzene are 100% and 93%, respectively, and the maximum concentrations are 57.03µg/L and 40.68 µg/L, respectively.The groundwater sampling point and the detection rate and pollution situation of each pollutant are listed in Figure S1 and Table S1 in Supplementary Materials.

FGPRB Packing and Structure Design
The FGPRB was composed of impervious curtains and permeable reactive barriers.The impervious curtain was 7 m long on both sides, and the permeable barriers were 10 m long in the middle, the width and depth of all walls were 1.5 m and 13 m, respectively, and the depth of embedded granite was approximately 1 m.Four monitoring wells (No.U1~U4) were set upstream of PRB, parallel to the reaction wall; two at the top of both wings of PRB, with a vertical distance of 4.0 m from the reaction wall, and 1.0 m from the edge of the impervious wall.The other two were 2.0 m upstream of PRB, with a distance of 4.0 m between them and 3.0 m away from the edge of the PRB.The U3 monitoring well was also used as a tracer injection well.There were three monitoring wells (No.R1~R3) inside the reaction wall, distributed equidistantly between the two layers of reaction walls.There were four monitoring wells (No.D1~D4) in the first row downstream, parallel to and 1.0 m away from the reaction wall.There was 1 monitoring well in the second row, 4.0 m away on the central vertical line of the PRB.The schematic diagram is shown in Figure 2. The anti-seepage curtain was made of cement and bentonite with a mass ratio of 1:1.It was injected into the ground through a sleeve valve grouting pipe.The permeability coefficient of the anti-seepage curtain was determined to be in the range of 5.2 × 10 −4 cm/s to 3.7 × 10 −5 cm/s by the water injection experiment.The permeable barrier was divided into two layers: the first layer was 0.47 m thick and composed of 50% zero-valent iron (ZVI) with a particle size of 3-5 mm, and 50% coarse sand with a particle size of 3-5 mm; the second layer was 0.93 m thick and composed of 25% activated carbon with a particle size of 3-5 mm, and 75% coarse sand with a particle size of 3-5 mm.

Numerical Simulation of Groundwater Barrier Wall
For the barrier wall, we used Visual MODFLOW for numerical simulation to clarify the delay time of pollutants in the downstream monitoring well due to different permeability and width.Two cases were designed, namely, point source pollution and non-point source pollution; the numerical simulation parameters are shown in Figure S2 and Table S2.The source intensity parameters of the point source pollution were: the number of pollutants per unit area, 100 g/m 2 ; leakage rate, 100 mm/y; leakage time, 365 d; point source area, 1 m 2 .The point source was located at the northern boundary of the parking lot in the original site, and a virtual monitoring well (B1) was set up 22 m downstream.The non-point source intensity parameters were: number of pollutants per unit area, 100 g/m 2 ; point source area, 9 m 2 , and the other parameters were the same as the point source pollution.The permeability coefficient of the PRB part was set as 7.7 m/d.This model only considers the adsorption of pollutants by the PRB medium, and its adsorption coefficient for benzene is 0.0034 mg/L.Water retaining walls with different parameters were set, and the simulation time was 7300 d.The specific parameters are listed in Table 2.

Groundwater Monitoring Methods
The earth excavation was carried out at the set position by a combination of excavator and manual work.After excavation, the grid skeleton was placed.The skeleton was composed of galvanized square pipe and stainless steel dense mesh, of which the outer skeleton was composed of 80 × 80 mm galvanized square pipe with a spacing of 200 mm.The stainless steel mesh was set outside the skeleton, with a wire thickness of 0.35 mm and a mesh size of 0.27 mm, to prevent filler overflow and impurity intrusion.After the completion of FGPRB, groundwater samples in different monitoring wells were collected regularly, the benzene and 1,2-dichloroethylene were detected as suggested by the STAN-DARD EXAMINATION METHODS FOR DRINKING WATER-ORGANIC PARAMETERS (GB/T 5750. . The tracer experiment was conducted by injecting 4 m 3 86 M NaCl solution into well U3.The change of electrical conductance (EC) at wells U2, R2, R3, D3 and D5 were monitored to analyze the influence of FGPRB on groundwater velocity and solute transport velocity.The location relationship of each monitoring well is shown in Figure 3.The on-line conductivity monitor was used to monitor the conductivity, and the monitoring range was 0-4000 µs/cm.

Benzene and 1,2-Dichloroethylene Removal by FGPRB
After completion of FGPRB, groundwater samples were planned for mid-month to clarify the concentration changes of pollutants in groundwater.Unfortunately, after the completion of FGPRB in April, affected by COVID-19, groundwater samples were not collected immediately.The concentrations of benzene and 1,2-dichloroethylene in different monitoring wells are shown in Table 3.The monitoring wells U1, U2, U3 and U4 are located upstream of FGPRB.The concentration of pollutants in each well varied little at different times, including the concentration of 1,2-dichloroethylene at 27.80-49.62µg/L, the pollution concentration of benzene was 22.35-31.75µg/L.However, the concentrations of 1,2-dichloroethylene and benzene in monitoring wells R1, R2 and R3 in FGPRB and downstream monitoring wells D1, D2, D3, D4 and D5 were low, and the detection data show that they were lower than the detection limit most of the time.This shows that FGPRB has a good effect on the remediation of 1,2-dichloroethylene and benzene in groundwater.In this field test, the PRB reaction system was filled with ZVI, activated carbon and coarse sand.It is speculated that when groundwater polluted by benzene and 1,2dichloroethylene pass through the PRB, the zero-valent iron material will dechlorinate and degrade 1,2-dichloroethylene [20] and have weak adsorption performance for benzene [21].When the groundwater penetrates the reaction zone of zero-valent iron, the remaining benzene and 1,2-dichloroethylene in groundwater will be adsorbed by activated carbon to further degrade benzene and 1,2-dichloroethylene in groundwater [22].In addition, the dissolved product Fe (III) of zero-valent iron can provide a reliable electron acceptor for indigenous microorganisms to degrade benzene and 1,2-dichloroethylene [23].However, these reaction processes are very complex.The filling material not only reacts with the target pollutants but also interacts with organic matter and various ions in groundwater.It needs to be further studied in combination with the long-term monitoring data and the changes of the filling material itself.

The Numerical Simulation Results of Groundwater Pollution Response Time
The pollution distribution after 365 d of operation of the model under different Cases is shown in Figure 4.When the FGPRB is not established, the pollutants will rapidly migrate downward in the direction of groundwater flow.The pollutants can be quickly monitored in monitoring well B1, and their migration speed is mainly controlled by dynamic groundwater conditions.After the FGPRB is established, the pollutants are controlled in the site, the pollution area of pollutants is reduced and the pollution concentration is also reduced at the same time.However, the effect of changing the thickness and permeability coefficient of the barrier wall on controlling pollutants is not obvious.This may be related to the hydrogeological conditions of the site itself.Similarly, under the condition of non-point source pollution, after the FGPRB is established, the pollutants are controlled in the site, the pollution area of pollutants is reduced and the pollution concentration is also reduced at the same time.Compared with point source pollution, the horizontal pollution range of the non-point source is slightly larger than that of the point source.Compared with point source pollution, the response time of pollutants in monitoring well B1 was basically the same after establishing FGPRB.However, the maximum concentration of pollutants in the monitoring well after establishing FGPRB was higher, which may be related to the longer residence time of pollutants in the monitoring well and the increase of pollutant accumulation.This is similar to the simulation results of Deng et al. [24] using numerical simulations to study the degradation of nitrate in groundwater before and after setting PRB.

The Changes of Monitor Well Electrical Conductance
Figure 6 shows the change of EC in monitoring well U2 with time, the EC has only a single peak response at 9:40 on April 14, rising from 987.2 µs/cm of the background value to 1791.9 µs/cm, and the end time of peak response is 19:53 on April 15.The conductivity of the U2 monitoring well began to change after adding tracer for 40 min, the period of maximum receiving conductivity was from 10:20 on April 14 to 13:50 on April 14.The EC value decreased from 3998.3 µs/cm to 1067.6 µs/cm within 60 min.Compared with the rising process of conductivity, the decline process of conductivity was relatively slow, and the conductivity continued to decline to the background value of 992.9 µs/cm, which took about 10 h.The reasons for the above phenomena may be as follows: (1) During drilling, the formation was disturbed, resulting in an increase of porosity upstream of the FGPRB reaction wall.At the same time, due to the construction of FGPRB, the groundwater velocity in the main runoff direction slowed down.That is, the permeability coefficient perpendicular to the main runoff direction of groundwater was greater than that of the main runoff direction of groundwater after the FGPRB was constructed.(2) The process of adding the tracer is equivalent to the water injection test.The increase of the hydraulic gradient led to an increase of groundwater velocity between injection well U3 and monitoring well U2, and the time of tracer receiving and declining was faster.(3) The main reason for the tailing phenomenon is due to the convection dispersion effect of the tracer.
At the same time, we set up monitoring wells R2 and R3 in FGPRB and also monitored the change of conductivity in R2 and R3 wells after injection of the tracer into the injection well U3; the monitoring results are shown in Figure 7.The conductivity monitoring data of monitoring well R2 changed significantly from 13:19 on May 7 to 13:41 on May 10, and the maximum conductivity was 1360 µs/m.It can be determined that the time from the injection well, adding tracer to the maximum conductivity of monitoring well R2, was 27 d, 3 h and 21 min.In addition, there were three obvious fast downward phenomena in the conductivity, which may be related to the three rainfall events in the test area.The variation trend of R3 in the monitoring well was similar to that of R2, but the value of conductivity fluctuates in the range of 648.93-97,933 µs/cm, and the variation value was similar to the background value, indicating that the tracer had not been received.
The change of conductivity in monitoring wells D3 and D5 downstream of FGPRB is shown in Figure 8.The conductivity values fluctuated up and down in the background value during the 48-day monitoring period; that is, the tracer was not received at U3 and U5.

The Changes of Groundwater Permeability Velocity after Construction of FGPRB
By comparing the response time of groundwater conductivity in monitoring wells U2 and R2, the influence of FGPRB on seepage velocity could be qualitatively analyzed.The monitoring well U2 is located at 0.5 • northwest of the injection well (U3), and the linear distance from the injection well is 2.42 m.The monitoring well R2 is located inside the reaction wall and it is 2.34 m away from the U3.However, the response time of conductivity of the two monitoring wells was quite different.The EC value of monitoring well U2 increased rapidly after 40 min of tracer injection while monitoring well R2 in the downstream FGPRB received the tracer only after 34,819 min.This phenomenon directly reflects that the permeability coefficient of the medium in the FGPRB is smaller than that in the site.Therefore, the existence of the PRB reaction wall significantly reduces the velocity of groundwater.
In order to further quantitatively analyze the influence of FGPRB on groundwater flow velocity, the monitoring data of monitoring well R2 were used to roughly calculate the average groundwater seepage velocity behind the FGPRB and the average permeability coefficient of water-bearing medium between injection well U3 and monitoring well R2 according to Darcy's law: where, S is the distance from injection well U3 to monitoring well R2; v is the average seepage velocity between the injection well U3 and the monitoring well R2; t is the time required for the tracer to be put into the monitoring well R2 to receive the maximum tracer concentration; K is the average permeability coefficient of the groundwater aquifer between injection well U3 and monitoring well R3; J is the hydraulic gradient.The calculation results are shown in Table 4.According to the previous pumping experiment, the average groundwater velocity between the injection well U3 and the monitoring well R2 is 0.078 m/d.According to the calculation, the average permeability coefficient of the groundwater aquifer between injection well U3 and monitoring well R2 is 31.2m/d.The permeability coefficient of the aquifer without FGPRB is 77 m/d, and the corresponding groundwater flow velocity is about 0.19 m/d.Therefore, the FGPRB constructed at the site can obviously slow down the groundwater seepage velocity.In order to further verify the influence of FGPRB on groundwater flow velocity, the theoretical time when the tracer was injected from injection well U3 reaches the maximum concentration after reaching the monitoring well R2 without FGPRB was obtained by Equations (3) and (4), and the calculation results were compared with the actual monitoring time after strong FGPRB response.The groundwater type in the study area is pore phreatic water in loose rock, which can be generalized as a one-dimensional groundwater flow field with infinite horizontal extension and equal thickness.The transport of pollutants in groundwater conforms to the two-dimensional phreatic hydrodynamic dispersion equation.Without considering the source-sink term, the mathematical model is shown in Equation (3).Hibsch and Krett et al. [25] have given the analytical solution of the mathematical model under the condition of two-dimensional hydrodynamic dispersion instantaneous injection in a onedimensional flow field, as shown in Equation (4).Where C is the concentration of NaCl in groundwater (g/L); C 0 is the background value of NaCl (g/L); x, y is the longitudinal and transverse migration distance of tracer (m); D L and D T are the longitudinal and transverse dispersion coefficients (m 2 /d), respectively; M is the injected amount of pollutants in the aquifer thickness (kg/m).Relevant parameters are shown in Table 5.The results show that the maximum concentration of the tracer from injection well U3 to monitoring well R2 is about 10 d without FGPRB; After the construction of FGPRB, the maximum concentration time from injection well U3 to monitoring well R2 was approximately 27 d, which further shows the water blocking effect of FGPRB.

Figure 3 .
Figure 3.The location relationship of each monitoring well.

Figure 4 .
Figure 4.The pollution distribution of 365 d was simulated under different cases.(a) No FGPRB is established (point sources); (b) Case 1; (c) Case 2; (d) Case 3; (e) Case 4;(f) No. FGPRB is established (non-point sources); (g) Case 5; (h) Case 6; (i) Case 7; (j) Case 8.In Figure 5, under the condition of point source pollution, the response time of target pollutants exceeding 0.01 mg/L in monitoring well B1 without the barrier wall was 25.91 d; after setting the barrier wall according to different scenarios, the response times of target pollutants exceeding 0.01 mg/L were 65.91 d, 65.0 d, 64.45 d and 64.45 d.The response time was significantly prolonged, but the pollution response time under different situations showed little difference.Under the condition of non-point source pollution, without FGPRB, the response time of pollutants was 27 d, and the response times of pollutants under different conditions were 66.51 d, 65.41 d, 65.41 d and 65.0 d.Compared with point source pollution, the response time of pollutants in monitoring well B1 was basically the same after establishing FGPRB.However, the maximum concentration of pollutants in the monitoring well after establishing FGPRB was higher, which may be related to the longer residence time of pollutants in the monitoring well and the increase of pollutant accumulation.This is similar to the simulation results of Deng et al.[24] using numerical simulations to study the degradation of nitrate in groundwater before and after setting PRB.

Figure 5 .
Figure 5.The target pollutant concentration in the monitoring well with time under different cases.

Figure 6 .
Figure 6.Change of EC in monitoring well U2.

Figure 7 .
Figure 7. Change of EC in monitoring wells R2 and R3.

Figure 8 .
Figure 8. Change of EC in monitoring wells D3 and D5.

Table 1 .
The stratum structure of the study area.

Table 2 .
List of setting parameters of partition wall.

Table 3 .
Change of pollutant concentration in each monitoring well.

Table 4 .
Parameter values and calculation results.