Numerical modelling and performance evaluation of multi-permeable reactive barrier system for aquifer remediation susceptible to chloride contamination

Journal Pre-proof Numerical modelling and performance evaluation of multi-permeable reactive barrier system for aquifer remediation susceptible to chloride contamination Rahul Singh, Sumedha Chakma, Volker Birke PII:

S2352-801X(19)30072-4

DOI:

https://doi.org/10.1016/j.gsd.2019.100317

Reference:

GSD 100317

To appear in:

Groundwater for Sustainable Development

Received Date: 1 March 2019 Revised Date:

18 October 2019

Accepted Date: 6 December 2019

Please cite this article as: Singh, R., Chakma, S., Birke, V., Numerical modelling and performance evaluation of multi-permeable reactive barrier system for aquifer remediation susceptible to chloride contamination, Groundwater for Sustainable Development (2020), doi: https://doi.org/10.1016/ j.gsd.2019.100317. This is a PDF file of an article that has undergone enhancements after acceptance, such as the addition of a cover page and metadata, and formatting for readability, but it is not yet the definitive version of record. This version will undergo additional copyediting, typesetting and review before it is published in its final form, but we are providing this version to give early visibility of the article. Please note that, during the production process, errors may be discovered which could affect the content, and all legal disclaimers that apply to the journal pertain. © 2019 Published by Elsevier B.V.

Performance evaluation of a Multi-Permeable Reactive Barrier system for contaminant plume treatment in the subsurface originating from multiple point sources using numerical modelling Contaminant Source 1

Contaminant Source 2

Residential Area

Partially treated water

Homogeneous Aquifer Continuous PRB 3

Contaminant Plume

Continuous PRB 1

Continuous PRB 2

Ground water flow

Treated water

1

Numerical Modelling and Performance Evaluation of Multi-Permeable

2

Reactive Barrier System for Aquifer Remediation Susceptible to Chloride

3

Contamination

4 5

Rahul Singh*

6

Research Scholar, Department of Civil Engineering, Indian Institute of Technology Delhi,

7

New Delhi, India: E-mail - [email protected]

8 9

Sumedha Chakma

10

Assistant Professor, Department of Civil Engineering, Indian Institute of Technology Delhi,

11

New Delhi, India: E-mail - [email protected]

12 13

Volker Birke

14

Professor, Mechanical Engineering / Process Engineering and Environmental Engineering

15

Hochschule Wismar, Germany: E-mail – [email protected]

16 17

*

18

Rahul Singh

19

E-mail - [email protected]

20

Telephone number - +49-17677896518

Corresponding author

21

1

22

Abstract Many in-situ groundwater remediation technologies have been developed, among

23

which the permeable reactive barrier (PRB) technology has emerged as an efficient, cost-

24

effective and sustainable remediation technique for the variety of contaminants. In this paper,

25

a numerical model is developed, using Visual MODFLOW, to evaluate the performance of a

26

multi-PRB system over the temporal and spatial groundwater quality variations. Model is

27

simulated for a single contaminant, i.e. Chloride (Cl-), released from multiple point sources,

28

over a hypothetical study area for a period of five years (1800 days). Initially, the model is

29

simulated without any remediation barrier and later multiple barriers, using Activated wood

30

charcoal (AWC) as a common reactive material, are introduced consecutively to contain the

31

plume to a desirable limit. Various parameters, such as the dimensions of the barriers and

32

continuous pumping, are taken into consideration for the performance evaluation of the

33

multi-PRB system. The results indicate that the performance of the multi-PRB system is more

34

efficient compared to the single PRB and natural attenuation system as the concentration in

35

all the wells could be seen drastically declined with the installation of PRBs. Thicker PRB

36

could produce better chloride removal rate due to the increase in residence time for the

37

adsorption of chloride over the reactive media. Further, the continuous pumping would also

38

increase the rate of remediation for the observation wells in its vicinity, however, up to a

39

certain limit. Furthermore, the maximum efficiency of the multi-PRB system can be achieved

40

at a lower depth compare to full study depth. Moreover, the PRBs adjacent to the

41

contaminant source treat the contaminants in the plume capture zone with high efficiency

42

than the far away PRBs. Finally, the numerical model shows that the contaminant plume,

43

containing chloride, is efficiently captured by the multi-PRB system in the proximity of the

44

point sources.

45

Keywords

46

MODFLOW; Activated wood charcoal; Continuous Pumping.

Groundwater

remediation;

Multi-Permeable

2

Reactive

Barrier;

Visual

47

1 Introduction

48

Recent years have witnessed an increasing concern over the deterioration of groundwater

49

quality due to various geogenic and anthropogenic sources like agricultural runoff, industrial

50

exertions, activated mine drainage, domestic and municipal solid wastes, etc. (Chakraborty et

51

al., 2010; Schipper et al., 2010; Wiafe et al., 2013; Rodak et al., 2014). These sources have

52

caused the emergence of numerous toxic and fatal contaminants in the groundwater like the

53

chlorinated compounds, hydrocarbons, heavy metals, etc., which have gathered worldwide

54

attention (Thiruvenkatachari et al., 2008; Obiro-Nayarko et al., 2014). More than 400,000

55

sites in the USA have been severely contaminated with toxic metals such as chlorinated

56

compounds and radioactive materials. Similarly, over Europe and Australia, there are well-

57

documented cases of groundwater pollution due to nitrates, hydrocarbons, chlorinated

58

compounds, sulfates, phosphates, etc. (Chakraborty et al., 2010; Thiruvenkatachari et al.,

59

2008; NRC, 1994). The increasing concentration of groundwater contamination has not only

60

led it unfit for drinking but also caused an adverse effect on humans, animals, and the

61

environment. (Suhag, 2016; Thiruvenkatachari et al., 2008). These severe incidents of

62

groundwater contamination create demand for the development of an efficient groundwater

63

remediation technique to eliminate the higher risk to health and the environment (Suhag,

64

2016).

65

Many conventional groundwater remediation techniques have been developed, over the past

66

few decades, among which the pump and treat (P&T) technology has been widely adopted

67

for the active remediation of contaminated groundwater above the ground surface (Phillips

68

and Atlas, 2005). The P&T technique is simple in application, however, it has low

69

remediation efficiency and is less cost-effective due to requirements of exhaustive energy and

70

input resources (USEPA, 2002). The drawbacks of the P&T technology have ushered the

71

evolution of many other ex-situ groundwater remediation technologies like steam stripping,

72

bioremediation, etc. (Thiruvenkatachari et al., 2008, ITRC, 2011).

73

technologies are also unable to restore the contaminated groundwater sites completely or

74

within the prescribed limit (USEPA, 1998). Further, many in-situ groundwater remediation

75

technologies have been developed, among which the Permeable Reactive Barrier (PRB) has

76

emerged as a promising alternative for the in-situ treatment of a variety of the groundwater

77

contaminants such as heavy metals, organics radionuclides, volatile organic carbons (VOCs),

3

However, these

78

chlorinated compounds, etc. (Thiruvenkatachari et al., 2008; Phillips, 2009; Xu et al., 2012;

79

Obiri-Nyarko et al., 2014; Faisal and Ali, 2017).

80

The PRB system primarily consists of a barrier filled with the reactive material which is

81

placed in the path of the contaminant plume at the subsurface to provide treated water

82

downstream from the barrier (Naidu and Birke 2014). The reactive materials, used in the PRB

83

system, play a very significant role to determine the performance efficiency of the entire

84

remediation process. The widely used reactive materials in PRB are Zero-valent iron (ZVI),

85

activated carbon (AC), limestone, activated alumina, sawdust, pea gravel, etc. The most

86

common reactive material among them is ZVI, which has been installed in more than 60% of

87

the PRBs in worldwide (ITRC, 2005). However, the treatment efficiency of ZVI is very

88

insignificant in many cases due to critical issues of precipitation and clogging at certain sites.

89

Therefore, AC has been applied widely for the removal of contaminants such as

90

hydrocarbons and chlorinated compounds (USEPA, 1998; Obiri-Nyarko et al., 2014). In

91

particular, for the treatment of chlorinated compounds, variety of AC have been used such as

92

commercial activated charcoal (CAC), activated coconut charcoal (ACC), heat-activated

93

bone charcoal (HABC) and activated wood charcoal (AWC) (Mukherjee et al., 2007;

94

Mohammed et al., 2012; Trubetskaya et al., 2019). AC and its different varieties provide

95

higher adsorption sites on their surface for the chloride and other contaminants, therefore,

96

many studies have shown AC as an effective and efficient adsorption material for various

97

organic pollutants (Webber and Morris, 1963; Mukherjee et al., 2007; Faust and Aly, 2013).

98

Adsorption of various pollutants over activated carbon is considered as the best available

99

technology (BAT), recommended by USEPA, among all the available technologies (Adams

100

and Watson, 1987). However, it is found that the CACs are very costly in nature and they

101

cannot use to treat a large range of chlorinated and other contaminants very efficiently for a

102

longer time period (Akl et al., 2014). Therefore, locally available material, i.e. AWC, as

103

steam activated biomass soot and tire carbon black, are more sustainable options for the

104

treatment of a large number of contaminants at a lower cost (Kennedy et al., 2004;

105

Mohammed et al., 2012; Trubetskaya et al., 2019).

106

The suitable design of the reactive barrier depends on many factors in which hydraulic

107

conductivities of the surrounding aquifer (Kaquifer) as well as the barrier (Kbarrier) are the most

108

significant. Kbarrier of the barrier must be higher than that of the Kaquifer to maximize the

109

remediation efficiency by allowing the contaminated groundwater plume to pass through the 4

110

barrier under a natural hydraulic gradient (Kacimov et al., 2011; Bortone et al., 2013).

111

Various studies have investigated the performance of the PRB system for the treatment of

112

different contaminants, on lab-scale experiments or large scale field assessment, (Blowes et

113

al., 1998; Vogan et al., 1999; Puls et al., 1999; Wilkin et al., 2003; Genç-Fuhrman et al.,

114

2007). However, prior to on-site PRB installation, the modelling of the PRB design would aid

115

in analyzing the performance of the PRB system for longer time periods and simulate its

116

behaviour under various plausible scenarios by varying factors such as the configuration of

117

PRB, location, and orientation of barrier, contaminant properties, etc.

118

Zeng and Wang (1999) and Konikow (2011) have used the numerical groundwater modelling

119

tools - MODFLOW and MT3DMS - for the flow and solute transport modelling respectively

120

to analyze the spreading of the contaminant plume in the aquifer system for a longer duration.

121

The MODFLOW and the MT3DMS tools work on finite difference techniques such as

122

explicit, implicit and Crank Nicolson finite difference methods. MODFLOW has been widely

123

used to quantitatively estimate the aquifer response concerning different input parameters

124

(McDonald and Harbaugh, 1988; Scott and Folkes, 2000; Zhan et al., 2009; Xu et al., 2012).

125

Scoot and Folkes (2000), Mayer et al. (2001) and Pandey and Mathur (2015) have introduced

126

reactive barriers in their groundwater models to analyze the behaviour of plume migration

127

through the PRB system. Further, few studies are carried out towards multiple reactive barrier

128

(multi-PRB) systems instead of a single barrier system for efficient groundwater remediation

129

(Birke et al., 2007, Lee et al., 2010; Xu et al., 2012).

130

A multi-PRB system, as shown in Fig. 1, is broadly defined as a sequence of PRBs placed

131

one-after-another and each PRB is either filled with the same or different reactive materials

132

based on the target contaminants (Birke et al., 2007). It can also be defined as a single barrier

133

filled with two or more reactive materials for the simultaneous removal of contaminants (Xu

134

et al., 2012; Obiri-Nyarko et al., 2014). Lee et al. (2010) recommended the use of numerical

135

modelling coupled with the multi-PRB system for the efficient remediation of a wide variety

136

and higher ranges of contaminants. Xu et al. (2012) have developed a model for a three-

137

dimensional flow and contaminant transport in an aquifer system subject to multi-PRB

138

remediation considering different forms of a single contaminant developed in various stages

139

of a chain reaction. However, they modelled the plume under the impact of a single

140

contaminant point source in a very small area and for a short time period. Birke et al. (2007)

141

have explained that treating a mixture of pollutants, without many difficulties, is easy with 5

142

the single reactive material in the multi-PRB system. Contrary, the situation is very difficult,

143

arises many problems such as clogging, uncertain interruption of fluid flow, etc., when

144

several reactive materials used sequentially in a multi-PRB system. Nevertheless, for the

145

higher range of single contaminant, it is very difficult to design and manage the hydraulics of

146

the system and its reactive behaviour with a large thickness of the barrier. The analysis of

147

these factors and their impact on the performance of the PRB system for a large area are rare

148

to find in the earlier studies. Moreover, the role of the location of the multiple reactive

149

barriers was unexplored in the modelling for the PRB system over a longer timeframe.

150

This paper aims to develop a multi-PRB numerical model using a modified MODFLOW

151

engine for the flow model followed by the MT3DMS engine for the contaminant transport

152

model. This numerical model aims to analyze the reactive efficiency of activated wood

153

charcoal (AWC), as a reactive barrier material, for the treatment of a single contaminant at

154

different stages, released from multiple point sources, in an artificially drawn study area. The

155

study is based on the advection-dispersion-reaction (ADR) equation considering the

156

adsorption as the reaction process for the groundwater remediation from all the PRBs in a

157

sequential manner. The three reactive barriers, in continuous configuration, are placed at

158

three different locations from the point sources. The VMOD is used as a numerical modelling

159

tool for the development of a flow and solute transport model for a multi-PRB system. The

160

MODFLOW-2000 and MT3DMSV5.2 are used as a flow and solute transport engine

161

respectively in VMOD. The three major aims of this paper are: a) to simulate the three

162

dimensional flow and solute transport model for the remediation of a homogeneous aquifer;

163

b) to analyse the removal efficiency of all the reactive barriers, under the impact of various

164

factors, over space and time for the treatment of contaminants released from multiple source;

165

c) to analyze the effect of continuous pumping on the removal efficiency of the PRB system.

166

2 Study Area

167

A hypothetical study area, as shown in Fig. 2, is considered for the development of the multi-

168

PRB numerical model to analyze the management of the contaminated plume. Mahar and

169

Data (2000) have introduced this study area for the identification of groundwater pollution

170

sources under transient conditions. Later, various researchers like Borah and Bhattacharjya

171

(2013, 2014), Singh et al. (2004), Srivastava and Singh (2015), Chaubey and Kashyap (2017)

6

172

have adopted the same study area for dealing with different aspects of groundwater

173

modelling.

174

The hypothetical study area covers a confined aquifer with an area of 1.04 km2 (1.3 km × 0.8

175

km) and a depth of up to 30 m. The details of the aquifer characteristics are shown in Table 1.

176

A continuous pumping well (P) is located at the center of the study area. The pumping rates

177

of the P, as shown in Table 2, are varied at an interval of 90 days (constant time step) with the

178

maximum and minimum pumping rates (m3/day) being 381.024 and 163.296 respectively.

179

The boundary conditions of the aquifer are considered to be time-invariant. The north and the

180

south boundaries are modelled as no-flow boundaries. The east and the west boundaries are

181

regarded as the constant head boundaries with the hydraulic head (m) varied from 100.00 to

182

99.58 and 88.00 to 87.72, respectively. Two potentially active contaminant sources (S1, S2)

183

are placed at the upgradient of the study area. The S1 and S2 sources are active for the first

184

360 days, with varying flux after every 90 days (as shown in Table 3). It is assumed that

185

Chloride (Cl-) is the only contaminant produced from all the active contaminant sources. The

186

eight observation wells, C1, C2, C3, C4, C5, C6, C7, and C8, are spread throughout the

187

aquifer. Aquifer parameters

Symbol

Hydraulic Conductivity in all three

Kxx

=

Kyy

=

Values Kzz

0.000191 m/s

dimensions

Effective Porosity

ᶯ

0.25

Longitudinal Dispersivity

DL

40.0 m

Transverse Dispersivity

DT

9.6 m

Specific Storage

S

0.002

188 189

Table 1. Aquifer characteristics for the study area.

190

Time step (90 days)

Pumping 3

rate (m /day)

Time step (90 days)

Pumping rate 3

(m /day)

7

Time step

Pumping rate

(90 days)

(m3/day)

1

273.024

8

273.024

15

381.024

2

163.296

9

381.024

16

217.728

3

327.456

10

217.728

17

273.024

4

163.296

11

163.296

18

163.296

5

273.024

12

327.456

19

327.456

6

327.456

13

273.024

20

217.728

7

163.296

14

163.296

-

191

Table 2. Pumping rate of water at pumping locations for different time steps (Borah and

192

Bhattacharjya, 2014).

193

Duration

Source flux (g/day)

(days)

S1

S2

S3

0-90

4,060,800

2,592,000

0

91-180

1,296,000

5,080,320

0

181-270

3,196,800

0

0

271-360

0

3,024,000

0

194 195

Table 3. Source Flux at different locations (Borah and Bhattacharjya, 2014).

196

197

3 Methodology

198

The general methodology adopted for modelling of PRB(s), is shown in Fig. 3 and briefly

199

described below. The first step is the identification of the target contaminant in the

200

considered study area. The second step is the development of flow and solute transport

201

modelling for the given aquifer system for the analysis of the contaminant plume spreading

202

throughout the area. Based on the identification of highly contaminated zones in the

203

subsequent step, the next step is the installation of a single or multiple PRBs. Further, the

204

PRB system is subjected to performance evaluation at every installation to check the

205

desirability of the PRB system for in-situ groundwater remediation from multiple

206

contaminant sources.

207

3.1 Assumptions 8

208

In this study, the following assumptions have been made: (a) the aquifer is considered to be

209

homogeneous; (b) the aquifer is considered to be isotropic (c) all contaminant sources are

210

producing one common contaminant, i.e. chloride (Cl-); (d) the initial concentration of Cl- in

211

all the wells is considered to be 100 mg/L; (e) the adsorption phenomenon is the only form of

212

remediation taken into consideration; (f) the Langmuir isotherm is adopted for the adsorption

213

phenomenon.

214

3.2 Governing Equations

215

The reactive transport of the solute in the aquifer was modelled using the advection-

216

dispersion-reaction (ADR) equation coupled with the flow equation. In tensor notation, the

217

three-dimensional transport of the solute can be represented by Equation (1) (Zheng and

218

Wang, 1999): 





    + =−  

±  





(1)



219

Where C

solute concentration (ML-3)

t

time (T)

ui

velocity in three dimensions (LT-1)

xi

longitudinal, lateral, and vertical distance (L)

Dij

dispersion coefficient tensor (L2T-1)

rm

physical, chemical, and biological reaction rates (ML-2T-1)

m

number of reaction types that have been taken place inside the barrier

220 221

The aquifer is considered to be homogeneous with the principal components of dispersion

222

represented by Dxx, Dyy, and Dzz for longitudinal, lateral, and vertical flow. Therefore,

223

Equation (1) can be simplified to a partial differential equation in three dimensions with

224

constant coefficients as shown by Equation (2).

225

 







 

 

 

= −[   +   +   ] + [    +    +    ] ± ∑  





(2)

226

The chemical interaction, specifically adsorption, of the solute with the surfaces of the porous

227

medium, in the PRB, can be defined by the following Equation (3) (Fetter, 1992):

9



228







0

229 230



+   =   ! − 

"# $

∑ &'() * − (+ ,- − (. ,

(3)

= () * − (+ ,

Where 12

bulk density (ML-3)

3

porosity of the porous medium (T)

()

adsorption rate coefficient

(+

desorption rate coefficient

S

sorbed concentration (ML-3)

(.

decay rate coefficient

*

a function that indicates the sorption isotherm

231 232

The *(C) function is mainly based on the best breakthrough curve which could follow either

233

the linear or the Freundlich or the Langmuir model.

234

It is assumed that adsorption was the only predominant process for the interaction of the

235

contaminant and the reactive materials inside the barrier. Therefore, Equation (3) can be

236

reduced to Equation (4) as shown below:

237

 







+   =   ! −

"# $

∑ &() *

(4)

238 239

3.3 Numerical Simulation of PRB

240

In this study, Visual MODFLOW version 2011.1 is used for the flow and transport simulation

241

in a homogeneous aquifer system. The entire hypothetical study area (1300 m × 800 m) is

242

disintegrated into smaller rectilinear grid cells of size 100 m × 100 m. A single layer of 30 m

243

depth is considered in all the computational grids of the model. After the definition of grid

244

size, various input parameters such as the hydraulic characteristics (conductivity,

245

transmissivity, specific yield, specific storage, effective porosity, and initial head), boundary

246

conditions (location of impermeable boundaries and constant heads) are initialized to the

247

model. Pumping well is also considered in the study, which constitutes an important

248

component of the solute transport in the aquifer system. The continuous pumping effectively

249

influences the water flow as well as solute transport within the aquifer and directly affect the 10

250

management of the groundwater system. Therefore, over-drafting of water, also termed as

251

continuous pumping, from the subsurface can unenviable changes in the movement of

252

contaminant transport in the aquifer system due to uncertain variation in head losses in the

253

aquifer system. Further, due to uncertain changes in the simulation of the flow and transport

254

model, the numerical simulation efficiency of the contaminant remediation from the reactive

255

barriers is also changed effectively. The centre of the study area is selected as the location of

256

the single continuous pumping to analyze the direct effect of solute transport in the whole

257

aquifer due to the continuous drafting of water from the aquifer as well as the effect on the

258

performance efficiency of the PRB system.

259

MODFLOW 2000, a three-dimensional finite-difference based groundwater flow modelling

260

tool, was used to simulate the flow under transient conditions. The contaminant transport

261

simulation was performed using MT3DMS, which followed the flow (MODFLOW)

262

simulation in order to analyze the contaminant spreading throughout the study area. The

263

MT3DMS tool was defined using some additional properties such as initial concentration and

264

two different point sources as the boundary conditions. The central finite difference method

265

under the implicit state was used to solve the advection-dispersion transport in the model.

266

The combined flow and contaminant transport model was run using WHS solver, which

267

works on a two-tier approach for the solution at each time step. Further, the flow and

268

transport model is simulated for 20-time steps in the entire five years’ period.

269

After a successful model run for the flow and contaminant transport modelling, the

270

installation of a single and multiple PRB systems is incorporated. The results of the

271

developed numerical model, using Visual MODFLOW, have been compared with the results

272

of the previous studies of Mahar and Dutta (2000), Borah and Bhattacharjya (2013 and 2014),

273

Srivastava and Singh (2015) performed on the same study area. The results could verify the

274

solute transport model including the temporal distribution of concentration in the area

275

covering all the observation wells. Thereafter, the continuous PRB configuration system,

276

followed by the developed flow and solute transport model, is adopted for the chloride

277

remediation produced from two different point sources. Initially, the 1st PRB is installed just

278

adjacent to the contaminant source (S1 andS2) followed by multi-PRB (2nd PRB and 3rd PRB)

279

installation, as shown in Fig. 2. The thickness of the PRBs is defined as 10 meters throughout

280

the depth of the barrier, i.e., 30 meters, of the study area. The PRBs are installed at an

281

orientation of 90◦ to the contaminant plume. Activated wood charcoal (AWC) is used as the

282

reactive material in the PRB system for the adsorption of the Cl- (Mohammed et al., 2012). 11

283

The adsorption parameters for the AWC reactive materials, such as adsorption capacity

284

(3.637 g/kg) and adsorption intensity (0.804 L/g) are obtained from Mohammed et al. (2012).

285

Many researchers have defined that, for a given PRB design, the discharge through the barrier

286

increases with an increase in Kbarrier (permeability of the barrier) relative to Kaquifer

287

(permeability of the aquifer) and vice versa (e.g., Gupta and Fox, 1999). Therefore, the

288

hydraulic conductivity of the reactive barrier is defined as three times that of the aquifer

289

hydraulic conductivity with a value of 5.7e-4 m/s. The 2nd PRB is installed close to the

290

contaminant source, however, at some distance to the 1st PRB. The 3rd PRB is installed far

291

away from the contaminant sources as well as from the 1st and 2nd PRBs.

292

The numerical simulation of the model is performed in six parts: (a) no PRB installation; (b)

293

after 1st PRB installation, (c) after 2nd PRB installation, (d) after 3rd PRB installation, (e) after

294

(1st+2nd) PRB installation together, (f) after (1st+2nd+3rd) PRB installation together.

295

4 Results and Discussions

296

The results of this study are divided into five major sections: (i) concentration analysis of the

297

wells, (ii) effect of PRB thickness variations, (iii) effect of pumping variations, (iv) effect of

298

depth variations, (v) spatial variation of the contaminant plume at different time-steps, (vi)

299

contaminant removal efficiency of the multi-PRB system.

300

4.1 Concentration Analysis of the Wells

301

The variation in the concentration of Chloride with time, in all the observation wells, for all

302

the cases of PRB installation is shown in Fig. 4. In case (a) with no PRB installation, the

303

observation wells, C2, C5, and C1, show the steep increase in the contaminant concentration

304

for the initial days in the observation period due to continuous emission of the contaminant

305

from the nearby contaminant sources, S1 and S2. Thereafter, the concentration in C2, C5, and

306

C1 reaches their respective maximum values and then subsequently decreases to zero until

307

the end of the observation period (1800 days). The well C5 shows a bi-modal concentration

308

profile, i.e., having two peaks, because of its proximity to the discontinuous source S2. The

309

continuation of the source (S2) for the first 180 days leads to the gradual increment of the

310

contaminant concentration in the well until the first peak is achieved. S2 is discontinued for

311

the next 90 days, leading to a decrease in concentration. After 270 days, S2 is re-continued

312

following which the concentration rapidly increases and attains the global peak. The other

313

wells C3, C4, C6, C7, and C8, which are located far away from source S1 and S2 show a very 12

314

minimal increase in contaminant concentration with respect to time. Moreover, they achieve

315

their peak concentration very late, 900 days after the source is deactivated. At the end of the

316

observation period, i.e., 1800 days, most of the wells, except C2, C5, and C1 are not

317

completely attenuated. The continuity-discontinuity of the contaminant sources near the

318

observation wells leads to the increment-decrement pattern in the concentration profile of

319

these wells. The concentration in all the observation wells declines mostly due to no

320

continuous external recharge of contaminants from both the sources as well as continuous

321

dilution due to pumping.

322

In case (b), i.e., 1st PRB installation, the similar increment-decrement pattern of the

323

concentration in the wells is observed, as shown in Fig. 4 (b). When compared to the case (a)

324

with no PRB installation, the concentration in all the observation wells drops threefold.

325

Similarly, the rate of increase in contaminant concentration is also decreased in all the

326

observation wells. This is evident as the time to reach the peak concentration is significantly

327

delayed. Moreover, the concentration profile of well C5, which is bimodal in case (a), shows

328

a single peak, as one of the peaks gets replaced by the point of inflection. It is noteworthy

329

here that the placement of the reactive barrier, in addition to remediating the groundwater,

330

retards the contaminant plume for a longer period when compared to the case (a) of the

331

absence of a barrier. At the end of the observation period, the wells C2, C5, and C1 show

332

higher concentrations, compared to all other observation wells due to retardation of the solute

333

transport upstream of all the observation wells. Therefore, the high intensity of the plume is

334

contained in the vicinity of the contaminant sources and these three observation wells, unlike

335

other far away observation wells.

336

Further, the installation of a 2nd PRB, upstream of all the observation wells, as shown in Fig.

337

4 (c) has been incorporated adjacent to the 1st PRB location. Both the reactive barriers have

338

shown almost similar results of concentration degradation with respect to time for all the

339

observation wells. However, in the case of wells C5 and C2, concentration degradation is

340

opposite for the entire observation period, 1800 days, in both the PRBs. As in the 1st PRB, the

341

rate of concentration degradation of well C5 is faster than well C2, however, the rate of

342

concentration degradation of well C2 is faster than well C5 after the installation of the 2nd

343

PRB. This manifests the different time period of concentration recharge in both the sources

344

and different time of interaction of the contaminant plume with the reactive barrier as the

345

location of both the barriers are different. Further, the 3rd PRB is placed downstream of

346

observation wells C2, C5, and C2 but upstream of all the other observation wells, i.e. C3, C4, 13

347

C5, C6, C7, and C8, as shown in Fig (4). Wells C2, C5, and C1 show worse results, as shown

348

in Fig. 4 (d), as compared to 1st PRB and 2nd PRB, however, the remaining wells show the

349

same results for all three PRBs. This clearly identifies the importance of the PRB location for

350

the remediation of a contaminated site.

351

After (1st+2nd) PRB (case (e)) and (1st+2nd+3rd) PRB (case (f)), simultaneous installation, the

352

concentration in most of the wells is further reduced, as compared to the individual

353

installation, as shown in Fig. 4 (e-f). However, due to the increased containment of the

354

contaminant plume by the (1st+2nd) and (1st+2nd+3rd) PRB, the rate of concentration decrease

355

in wells after discontinuity of all sources is significantly lowered. With the installation of

356

(1st+2nd+3rd) PRB, a concentration reduction in wells C3, C4, C6, C7, and C8 is observed

357

when compared to their concentrations during the (1st+2nd) PRB installation. However, no

358

such reduction in the concentration is observed for the observation wells lying upstream of

359

the 3rd PRB, i.e., C2, C5, and C1.

360

The installation of the (1st+2nd+3rd) PRB does not show a significant reduction in the peak

361

concentration of the contaminant. The concentration in C2, C5, and C1 is still higher than the

362

permissible limit of Cl-. Since the third PRB is placed far away from the source and therefore,

363

it lies in the downstream of the observation wells C2, C5, and C1, it plays no role in reducing

364

their contamination. This raises a critical point that the location of PRB plays a vital role in

365

increasing the remediation efficiency of the entire aquifer system. The PRBs which placed

366

nearby the pollution sources are more likely to remediate the aquifer compared to PRBs

367

which are far away from the source.

368

The peak contaminant concentration in all the wells, for all the four cases of the PRB system,

369

is shown in Fig. 5. The installation of the multi-PRB system sequentially reduces the

370

contaminant concentration in all the wells. In case (a) with no PRB installation, the maximum

371

peak concentration of the contaminant is observed in C2 followed by C5 and C1 due to their

372

proximity to the contaminant source. However, the other wells, i.e., C6, C3, and C7 show

373

relatively lower concentration. The minimum peak concentration of the contaminant is

374

observed in C8 and C4. These results reveal that the natural attenuation system of the aquifer

375

is not sufficient to remediate the contaminant up to the required plausible limit (permissible

376

limit of chloride in drinking water). Thereafter, the installation of the 1st PRB led to a

377

decrease in concentration at C2 and C5 along with other observation wells. Further, with the

378

installation of the (1st+2nd) and (1st+2nd+3rd) PRB, together, the concentration in C2, C5, and

14

379

C1 decreases more compared to individual installations of 1st PRB, 2nd PRB, and 3rd PRB.

380

However, the concentration in C5 drops down drastically when compared to C1 and C2

381

because C5 is closer to the continuous pumping location (P) than to C1 and C2. Further, the

382

concentration in C3, C4, C6, C7, and C8 observation wells drops below the required plausible

383

limit.

384

4.2 Effect of PRB Thickness Variations

385

The effect of various thicknesses of reactive barrier in a single PRB system for chloride

386

degradation, releasing from the multiple contaminant sources, is represented in Fig. 6. The

387

barrier thickness has been varied from 1 meter to 10 meters. It is clearly shown that the

388

thicker PRB shows better chloride degradation results than the thinner reactive barrier. The

389

thickest PRB, i.e., 10 meters, manifests the highest drop in chloride concentration for the

390

nearest observation wells C2 and C5 from the source (below 1000 mg/l), as shown in Fig. 6

391

(f). However, the thinnest PRB, i.e., 1 meter, manifests a small decline in chloride

392

concentration but the concentration is found more than 2000 mg/l, as shown in Fig. 6 (a), for

393

the same observation wells. The other cases of barrier thickness show the chloride

394

degradation results in between thickest, i.e., 10 meters, and thinnest PRB, i.e., 1 meter. It can

395

be clearly observed that the more residence time has been provided to the contaminant plume

396

for the thick barrier, which leads the contaminant plume for better adsorption process over

397

the barrier’s reactive media, compared to the thin PRB.

398

4.3 Effect of Continuous Pumping Variations

399

Three different cases of pumping variations have been observed when (a) the rate of

400

continuous pumping is P, as mentioned in Table 2., (b) the rate of continuous pumping is five

401

times more than the current pumping i.e., 5P, and (c) the rate of continuous pumping is ten

402

times more than the current pumping i.e., 10P. The effect of continuous pumping variations

403

has been observed for four different installations of PRBs i.e. without PRB, 1st PRB, (1st+2nd)

404

PRB and (1st+2nd+3rd) PRB as shown in Fig. 7. It has been observed that the increase in

405

continuous pumping rate from P to 5P leads to dilution of contaminant concentration in all

406

the observation wells. The high rate of dilution has been observed in the without PRB case,

407

Fig. 7 (a & e), however in all other PRB cases, Fig. 7 (b & f) – 1st PRB, Fig. 7 (c & g) -

408

(1st+2nd) PRB, and Fig. 7 (d & h) (1st+2nd+3rd) PRB, the rate of dilution of concentration is

409

lower due to contaminant retardation within the reactive barriers. Further, the rate of pumping

410

has been increased to 10P which also displays almost similar results, Fig. 7 - (i), (j), (k) & (l), 15

411

or the same rate of dilution in contaminant concentration as in the case of 5P. This exhibits

412

the saturation of dilution rate of contaminant concentration at 5P pumping rate and thereafter

413

the changes in the concentration are very low for all the observation wells.

414

4.4 Effect of Depth Variations

415

The multiple PRBs performances have been analyzed for four different depth variations

416

between 0-30 meters. Three different PRB cases, i.e., 1st PRB, (1st PRB+2nd PRB) and (1st

417

PRB+2nd PRB+3rd PRB), have been taken for performance analysis for four different depths,

418

5 meters, 10 meters, 20 meters, and 30 meters. In this study, all the scenarios have been

419

analyzed at a depth of 30 meters. However, in three-dimensional modelling, it is significant

420

to analyze the reactive barrier performance at different depths in the porous media. Therefore,

421

the comprehensive analysis for all the observation wells has been shown in Fig. 8 as

422

concentration vs time graph. The degradation of contaminant concentration is very low at 5

423

meters and 10 meters depths for all the multi PRB installation as shown in Fig. 8 (a-f).

424

However, the pattern of rising and fall of concentration vs time graph for all the observation

425

wells is similar to the without PRB installation case as shown in Fig. 4 (a). This clearly

426

displays the very small rate of contaminant degradation at these depths, therefore, for high

427

degradation rates PRB must be installed at higher depth at this particular study area.

428

Further, the installation of the PRB at 20 meters’ depth has significantly changed the rate of

429

contaminant degradation compared to the lower depth as shown in Fig. 8 (g-i). The

430

contaminant concentration has come down below the required plausible limit except at

431

observation wells C2, C5 and C1, however, these wells also show a significant drop in the

432

concertation at this depth as shown in Fig. 8 (g) for 1st PRB case. At the same depth, the rate

433

of degradation has increased after the installation of multi PRBs (1st PRB+2nd PRB) together,

434

as shown in Fig 8 (h). The same concentration drop has been displayed by (1st PRB+2nd

435

PRB+3rd PRB) installation simultaneously, as shown in Fig. 8 (i).

436

However, wells downstream of 3rd PRB have shown more significant changes after the

437

installation of three PRBs together compared to upstream wells. This is because the

438

contaminant plume is treated thrice before reaching downstream wells compared twice for the

439

upstream wells. The rate of contaminant degradation, for all the observation wells, at a depth

440

of 30 meters is similar to the rate of degradation at 20 meters, as shown in Fig. 8 (j-l), for all

441

the PRB installations. It is evident from this depth variation study that the maximum removal

442

efficiency is achieved at a depth of 20 meters. Therefore, this study reveals that instead of 16

443

going for the full depth of 30 meters, the maximum efficiency of a multi reactive barrier

444

system for this hypothetical example can be achieved at a lower depth of 20 meters.

445

4.5 Spatial Variation of Contaminant Plume at Different Time-steps

446

The spatial variation of the contaminant plume in the entire aquifer is analyzed at four

447

different time steps - 90 days, 450 days, 900 days and 1800 days for each PRB installation

448

case, as shown in Fig. 9-12.

449

Fig. 9 shows the spatial distribution of contaminant plume in the case of no PRB installation.

450

It is observed that with the passage of time, the plume spreads and moves away from the

451

point source. The concentration intensity of the plume is at a maximum near the source and

452

radially decreases while moving away from the source. At the end of 450 days, the plume

453

spread almost reaches the center of the aquifer (as shown in Fig. 9 (b)). The plume spreads in

454

an oval shape which indicates that the longitudinal dispersivity (X direction) is higher than

455

transverse dispersivity (Y direction) within the aquifer. At the end of 900 days, the plume

456

spread increases significantly, however, the strength of the plume is reduced. Due to lower

457

dispersivity in the transverse direction, the plumes from sources S1 and S2 overlap as shown

458

in Fig. 9 (c) and their cumulative spread encompasses all the observation wells. It is also

459

observed that after the discontinuity of source S1 and S2, the contaminant plume moves away

460

from their source profoundly in the downstream direction. At the end of the simulation

461

period, i.e., 1800 days the overlapped plume has entirely left the contaminant source and

462

moved to the downgradient half of the aquifer. Moreover, the maximum concentration

463

intensity of the plume is significantly reduced to 750 mg/L as shown in Fig. 9 (d).

464

The installation of the 1st PRB significantly decreases the contaminant intensity of the plume

465

at all the time steps as shown in Fig. 10. However, the plume spread of plume is reduced at

466

the initial time-steps only. Due to the placement of the 1st barrier just adjacent to the sources,

467

the management of plume near the source takes place; as a result, the plume is unable to leave

468

its originating point sources even after many days of discontinuity of the source. Therefore,

469

when compared to the case of no PRB installation, the longitudinal plume spread is higher

470

within the considered section of the aquifer. Nevertheless, the 1st PRB is highly effective in

471

minimizing the overall spread and intensity of contamination in the considered aquifer

472

system.

473

The spreading of the plume and the contaminant intensity further reduces by installing the

474

(1st+ 2nd) PRB, at a small distance away from the sources (as shown in Fig. 11). Similar to the 17

475

previous case, the plume does not leave the discontinuous source until the end of the

476

observation period. The PRBs not only act as barriers that treat the water passing through it

477

but also contain the contaminant for a longer time upstream of the barrier. The installation of

478

the (1st+2nd+3rd) PRB where 3rd PRB has been installed away from the contaminant source

479

further reduces the spread and the intensity of contamination in the aquifer as shown in Fig.

480

12. The concentration of contaminant decreases below the required permissible limit of Cl-

481

within the entire aquifer during the observation period of 1800 days. Moreover, the plume is

482

completely contained between the source and the third barrier, thus reducing the spread of the

483

contaminated plume significantly at the end of 1800 days. This shows that the water passing

484

through the third barrier is completely remediated.

485

4.6 Contaminant Removal Efficiency of the Multi-PRB System

486

The contaminant removal efficiency of the multi-PRB system has been defined by its

487

percentage of contaminant concentration degradation in all the observation wells after every

488

PRB installation in the aquifer. The removal efficiency is defined by equation (5). % 56789 =

489

::;< = ;>

;<

(5)

490

Where, ? is the concentration of the ith well in mg/l without any PRB installation

491

throughout the aquifer;  is the concentration of the ith well in mg/L after jth PRB

492

installation.

493

The removal efficiency of the multi-PRB system is evaluated with reference to the no PRB

494

installation case. Fig. 13 shows the percentage removal of the contaminant in each

495

observation well after the installation of the 1st PRB, 2nd PRB, 3rd PRB, (1st+2nd) PRB and

496

(1st+2nd+3rd) PRB. The contaminant removal starts early in wells C2, C5 and C1 as compared

497

to other observation wells (as shown in Fig 13 (b, e, a)) as the treated plume from the 1st and

498

2nd PRB reach them early when compared to the other observation wells. However, the

499

percentage removal curve of the 3rd PRB is shown to be the lowest removal efficiency for

500

these observation wells. This is because the 1st PRB and 2nd PRB have been installed at the

501

upstream of all the observation wells but the 3rd PRB has been installed downstream of C2,

502

C5 and C2 and upstream of the other wells. Therefore, the contaminant retardation has not

503

occurred in these upstream wells due to the 3rd PRB installation. Nevertheless, the percentage

504

removal curves of the (1st+2nd) and (1st+2nd+3rd) PRB is overlapped for these wells

505

throughout the observation period because these wells are located in the downstream of 1st 18

506

and 2nd PRB but upstream of the 3rd PRB, therefore, no major changes have occurred in these

507

wells after the installation of 3rd PRB. However, in the case of the other wells (C3, C4, C6,

508

C7, and C8), situated downstream from the 3rd PRB, the overlap of the removal curves takes

509

place for the first half of the simulation period, i.e., 900 days. However, wells C3 and C7

510

show the incomplete contaminant remediation after all the three PRB installation, at the end

511

of the observation time (1800 days). Due to the lower dispersivity of the aquifer in the

512

transverse direction and the nonalignment of these wells with the contaminant source, the

513

treated plume reaches these wells very late.

514

It is observed that the percentage removal of the contaminants sequentially increased with the

515

installation of multiple numbers of PRBs. The contaminated plume initially gets treated with

516

the first PRB, followed by the second PRB and third PRB sequentially. Therefore, the

517

efficiency of the (1st+2nd+3rd)

518

PRB. These results indicate the higher efficiency of the multi-PRB compared to single PRB

519

in remediating the aquifer efficiently.

PRB is higher than the (1st+2nd) PRB followed by the 1st

520 521

5 Conclusions

522

This paper has investigated the in-situ remediation of a homogeneous aquifer system with a

523

single contaminant by numerical modelling of a multi-PRB system. A hypothetical study

524

area, having two active point sources of contamination and continuous pumping, is modelled.

525

MODFLOW is used to simulate the flow model followed by MT3DMS for the solute

526

transport model. Three PRBs, continuously configured are installed at three different

527

locations from the source. The performance of the multi-PRB system is evaluated for a time

528

period of five years (1800 days).

529

It is observed that the performance of multiple PRBs is superior to a single PRB system and

530

natural attenuation in the aquifer processes. In the case of no PRB system, the maximum

531

intensity of contaminant plume is significantly higher than the permissible limit as the

532

concentration in all the observation wells are many folds higher than the desired standard

533

limits. Therefore, with the installation of the 1st PRB, adjacent to the contaminant sources, the

534

plume concentration intensity dropped down remarkably so as the concentration in all the

535

observation wells. However, the 1st PRB installation is still unable to achieve the reduction up

536

to the permissible contaminant limit. Consequently, the second and third PRBs are installed 19

537

simultaneously to remediate the aquifer to achieve the essential limit of the contaminant in

538

the groundwater. Since, the concentration in all, near and far away, wells are dropped to a

539

required value.

540

The thickness of the PRB is a guiding factor in determining the efficiency of PRB in the

541

remediation of the contaminant plume. The thick barriers can be provided more residence

542

time to the contaminant plume for better reactivity with the reactive material compare to the

543

thin PRB. Multiple PRBs, with the limited barrier thickness, also drop down the chances of

544

frequent hydraulic loss and help to manage the reactive behaviour between the contaminant

545

and reactive material, when installed on the field. In addition, the continuous pumping also

546

increases the remediation rate of the multi-PRB system for the area in its vicinity. Further, it

547

is observed that the depth of reactive barrier installation is also important to analyze, before

548

field installation, to achieve the optimum depth of the barrier for obtaining maximum

549

removal efficiency. The depth variation in this study showed that the maximum contaminant

550

degradation can be achieved at a depth less than the full depth possible for the study.

551

Therefore, on the field study, the installation of reactive barriers at the right depth should lead

552

to a cost-saving by minimizing the size and number of required reactive barriers. Further, it is

553

also observed that the PRBs which are in the proximity of contaminant source contain the

554

plume more significantly than the far away PRBs.

555

Furthermore, it is observed that dispersion plays an important role in governing the treatment

556

rate of the contaminant plume for a long period of time. The low transverse dispersion

557

coefficient results in a lower treatment rate of the plume that reaches the observation wells

558

are not aligned with the point source. On the other hand, the high longitudinal dispersion

559

leads to a higher rate of plume treatment to the observation wells aligned with the

560

contaminant source. Moreover, the results suggest that the performance evaluation of multi-

561

PRB systems is very significant through numerical modelling for a longer duration.

562

Acknowledgment

563

This research is fully supported by the Indian Institute of Technology Delhi, India. We would

564

like to thank Mr. Shushobhit Chaudhary from the Indian Institute of Technology Delhi for his

565

insight and expertise on the subject that greatly assisted this research work.

566

References 20

567

Adams, C.D., and Watson, T.L. (1996). Treatability of s-triazines herbicides metabolites

568

using powdered activated carbon. Journal of Environmental Engineering, 122 (4),

569

327-330.

570

Akl, M.A., Dawy, M.B., Serage, A.A. (2014). Efficient removal of phenol from water

571

samples using sugarcane bagasse based activated carbon. Journal of Anal Bioanal

572

Technology, 5 (2).

573

Birke V.O., Burmeier H.A., Jefferis S.T., Gaboriau H.E., Touze S., Romain C. (2007).

574

Permeable reactive barriers (PRBs) in Europe: potentials and expectations. Italian

575

Journal of Engineering Geology and Environment, 1:1-8.

576

Blowes D.W., Ptacek C.J., Benner S.G., McRae C.W., Puls R.W. (1998). Treatment of

577

dissolved metals using permeable reactive barriers. IAHS Publication (International

578

Association of Hydrological Sciences), 483-90.

579

Borah T., Bhattacharjya R.K. (2014). Development of Unknown Pollution Source

580

Identification

581

Methodology. Journal of Hazardous, Toxic, and Radioactive Waste, 19(3): 04014034.

582

Borah T., Bhattacharjya R.K. (2013). Solution of source identification problem by using

583

Models

Using

GMS

ANN–Based

Simulation

Optimization

GMS and MATLAB. ISH Journal of Hydraulic Engineering, 19(3):297-304.

584

Bortone I., Chianesea S., Di Nardoa A., Di M., Natalea A.E., Musmarraa D. (2013). A

585

Comparison between pump & treat technique and permeable reactive barriers for the

586

remediation

587

compounds. Chemical Engineering, 32.

588 589

of

groundwater

contaminated

by

chlorinated

organic

Chakraborti D., Das B., Murrill M.T. (2010). Examining India’s groundwater quality management. Environmental Science Technology, 45:27-33.

590

Chaubey J., Kashyap D. (2017). A data parsimonious model for capturing snapshots of

591

groundwater pollution sources. Journal of Contaminant Hydrology, 197:17-28.

592

Faisal A.A.H., Ali Z.T.A. (2017). Using sewage sludge as a permeable reactive barrier for

593

remediation of groundwater contaminated with lead and phenol. Separation Science

594

and Technology, 52(4):732-742.

595

Faust, S. D., Aly, O. M. (2013). Adsorption processes for water treatment. Elsevier.

596

Fetter C.W. (1992). Contaminant Hydrogeology, second ed. Prentice-Hall, Englewood Cliffs,

597

NJ, p458.

21

598

Genç-Fuhrman H., Mikkelsen P.S., Ledin A. (2007). Simultaneous removal of As, Cd, Cr,

599

Cu, Ni and Zn from stormwater: Experimental comparison of 11 different sorbents.

600

Water Research, 41(3):591-602.

601 602

Gupta, N., Fox, T. C. (1999). Hydrogeologic modeling for permeable reactive barriers. Journal of Hazardous Materials, 68 (1-2), 19-39.

603

Interstate Technology & Regulatory Council (ITRC) (2011). Permeable Reactive Barrier:

604

Technology Update. PRB: Technology Update Team, Washington, D.C., 156.

605

Interstate Technology Regulatory Council (ITRC) (2005). Permeable Reactive Barriers:

606

Lessons Learned/New Directions. PRB: Technology Update Team, Washington, DC,

607

101.

608

Kacimov A.R., Klammler H., Il’yinskii N., Hatfield K. (2011). Constructal design of

609

permeable reactive barriers: groundwater-hydraulics criteria. Journal of Engineering

610

Mathematics, 71 (4):319–338. http://doi.org/ 10.1007/s10665-011-9457-5

611

Kennedy, L.J., Vijaya, J.J., Sekaran, G. (2004). Husk by phosphoric acid activation. Effect of

612

two-stage process on the preparation and characterization of porous. Industrial &

613

engineering chemistry research, 43 (8), 1832-1838.

614

Konikow

L.F.

(2011).

The

Secret

to

Successful

Solute‐Transport

615

Modeling. Groundwater, 49(2):144-159. Lee J.Y., Lee K.J., Youm S.Y., Lee M.R.,

616

Kamala-Kannan S., Oh B.T. (2010). Stability of multi-permeable reactive barriers for

617

long term removal of mixed contaminants. Bulletin of Environmental Contamination

618

and Toxicology, 84(2):250-254.

619 620

Mahar P.S., Datta B. (2000). Identification of pollution sources in transient groundwater systems. Water Resources Management, 14(3):209-227.

621

Mayer K.U., Blowes D.W., Frind E.O. (2001). Reactive transport modeling of an in situ

622

reactive barrier for the treatment of hexavalent chromium and trichloroethylene in

623

groundwater. Water Resources Research, 37(12):3091-103.

624 625

McDonald M.G., Harbaugh A.W. (1988). A modular three-dimensional finite-difference ground-water flow model. Reston, VA: US Geological Survey, 6:A1.

626

Mohammed I., Ariahu C.C., Nkpa N.N., Igbabul B.D. (2012). Chlorine adsorption kinetics of

627

activated carbon from selected local raw materials. Journal of Chemical Engineering

628

and Materials Science, 3(2), 23-29.

629

Mukherjee, S., Kumar, S., Misra, A. K., & Fan, M. (2007). Removal of phenols from water

630

environment by activated carbon, bagasse ash and wood charcoal. Chemical

631

Engineering Journal, 129(1-3), 133-142. 22

632 633 634 635

Naidu R., Birke V. (Eds.) (2014). Permeable reactive barrier: sustainable groundwater remediation. CRC Press, 1. National Research Council (NRC) (1994). Ground water recharge using waters of impaired quality. National Academies Press, Washington, DC, pp382.

636

Pandey M.P., Mathur S. (2015). Mathematical Modeling and Application of Permeable

637

Reactive Barrier into a Hazardous Waste Contaminated Site. Journal of Advances in

638

Engineering Science and Management, 4(3):232-238.

639 640

Phillips D.H. (2009). Permeable reactive barriers: A sustainable technology for cleaning contaminated groundwater in developing countries. Desalination, 248(1-3):352-359.

641

Philip J., Atlas R.M. (2005). Bioremediation of contaminated soils and aquifers. Applied

642

Microbial Solutions for Real-World Environmental Cleanup. ASM Press, Washington

643

D.C., pp139-236.

644

Puls R.W., Blowes D.W., Gillham R.W. (1999). Long-term performance monitoring for a

645

permeable reactive barrier at the US Coast Guard Support Center, Elizabeth City,

646

North Carolina. Journal of Hazardous Materials, 68(1-2):109-24.

647

Rodak C., Silliman S.E., Bolster D. (2014). Time‐Dependent Health Risk from Contaminated

648

Groundwater Including Use of Reliability, Resilience, and Vulnerability as Measures.

649

JAWRA Journal of the American Water Resources Association, 50(1): 14-28.

650

Obiri-Nyarko F., Grajales-Mesa S.J., Malina G. (2014). An overview of permeable reactive

651

barriers for in situ sustainable groundwater remediation. Chemosphere, 111:243-259.

652

Schipper L.A., Robertson W.D., Gold A.J., Jaynes D.B., Cameron S.C. (2010). Denitrifying

653

bioreactors—an approach for reducing nitrate loads to receiving waters. Ecological

654

Engineering, 36(11):1532-1543.

655

Scott K.C., Folkes D.J. (2000). Groundwater modeling of a permeable reactive barrier to

656

enhance system performance. Proceedings of the 2000 Conference on Hazardous

657

Waste Research, Denver, Colorado, USA.

658

Singh R.M., Datta B., Jain A. (2004). Identification of unknown groundwater pollution

659

sources using artificial neural networks. Journal of Water Resources Planning and

660

Management, 130(6):506-514.

661

Srivastava D., Singh R.M. (2015). Groundwater system modeling for simultaneous

662

identification

663

characterization. Water Resources Management, 29(13):4607-4627.

664 665

of

pollution

sources

and

parameters

with

uncertainty

Suhag R. (2016). Overview of Ground Water in India. PRS Legislative Research Standing Committee on Water Resources: Delhi, India, (No. id: 9504). 23

666

Thiruvenkatachari R., Vigneswaran S., Naidu R. (2008). Permeable reactive barrier for

667

groundwater remediation. Journal of Industrial and Engineering Chemistry, 14(2),

668

145-156.

669

Trubetskaya, A., Kling, J., Ershag, O., Attard, T. M., Schröder, E. (2019). Removal of phenol

670

and chlorine from wastewater using steam activated biomass soot and tire carbon

671

black. Journal of hazardous materials, 365, 846-856.

672

USEPA (1998). Evaluation of Demonstrated and Emerging Technologies for the Treatment

673

of Contaminated Land and Groundwater (Phase III) Treatment Walls and Permeable

674

Reactive Barriers. EPA 542-R-98-003. U.S. Environmental Protection Agency,

675

Vienna, Austria, pp101.

676

USEPA (2002). Field Applications of In Situ Remediation Technologies: Permeable Reactive

677

Barriers. U.S. Environmental Protection Agency, Office of Solid Waste and

678

Emergency Response, Washington, DC.

679

Vogan J.L., Focht R.M., Clark D.K., Graham S.L. (1999). Performance evaluation of a

680

permeable reactive barrier for remediation of dissolved chlorinated solvents in

681

groundwater. Journal of Hazardous Materials, 68(1-2):97-108.

682 683 684 685 686 687 688

Weber W.J., and Morris, J.C. (1963). Kinetics of adsorption on carbon from solution. Journal of the Sanitary Engineering Division, 89 (2), 31-60. W.H.O. (2011). Guidelines for drinking-water quality. WHO Chronicle, Edition F, 38(4):104-108. Wiafe G., Boateng I., Appeaning-Addo K. (2013). Handbook for coastal processes and management in Ghana. Choir Press. Wilkin R.T., Puls R.W., Sewell G.W. (2003). Long‐term performance of permeable reactive

689

barriers

using

zero‐valent

690

Groundwater, 41(4):493-503.

iron:

Geochemical

and

microbiological

effects.

691

Xu Z., Wu Y., Yu F. (2012). A three-dimensional flow and transport modeling of an aquifer

692

contaminated by perchloroethylene subject to multi-PRB remediation. Transport in

693

Porous Media, 91(1):319-337.

694

Zheng C., Wang P.P. (1999). MT3DMS: a modular three-dimensional multispecies transport

695

model for simulation of advection, dispersion, and chemical reactions of contaminants

696

in groundwater systems; documentation and user's guide. Alabama University.

697

Zhan H., Wen Z., Gao G. (2009). An analytical solution of two‐dimensional reactive solute

698

transport in an aquifer‐aquitard system. Water Resources Research, 45(10).

24

Figure Captions Fig. 1 In-situ multi-PRB groundwater remediation system.

Fig. 2 Schematic representation of the hypothetical study area with (a) Plan view and (b) elevation view of 1st, 2nd and 3rd installed PRB. Fig. 3 Comprehensive methodology of the Permeable Reactive Barrier (PRB) modelling for groundwater contaminant(s) removal. Fig. 4 Variation of Chloride concentration with time in all observation wells for the cases of (a) no PRB installation, (b) 1st PRB installation, (c) 2nd PRB installation, (d) 3rd PRB installation, (e) (1st+2nd) PRB installation, and (f) (1st + 2nd + 3rd) PRB installation. Fig. 5 Maximum concentration of contaminants in all the monitoring wells in various cases of PRB installation.

Fig. 6 Concentration vs time graph for single PRB for six different cases of barrier thickness i.e. (a) 1 meter, (b) 2 meters, (c) 4 meters, (d) 6 meters, (e) 8 meters, and (f) 10 meters. Fig. 7 Concentration vs time graph for all the cases of PRB i.e. without PRB, 1st PRB, (1st+2nd) PRB and (1st+2nd+3rd) PRB for the variation in continuous pumping rate (a-d) Pumping rate = P; (e-h) Pumping rate = 5P; (i-l) Pumping rate = 10P respectively. Fig. 8. Concentration vs time graph of 1st PRB, (1st+2nd) PRB and (1st+2nd+3rd) PRB at a depth of (a-c) 5 meters, (d-f) 10 meters, (g-i) 20 meters and (j-l) 30 meters respectively.

Fig. 9 Spatial variation of contaminant plume (mg/L) in case of no PRB installation at time steps of (a) 90 days, (b) 450 days, (c) 900 days and (d) 1800 days.

31

Fig. 10 Spatial variation of a contaminant plume in case of 1st PRB installation at time steps of (a) 90 days, (b) 450 days, (c) 900 days and (d) 1800 days. Fig. 11 Spatial variation of a contaminant plume in case of (1st+2nd) PRB installation at time steps of (a) 90 days, (b) 450 days, (c) 900 days and (d) 1800 days. Fig. 12 Spatial variation of a contaminant plume in case of (1st+2nd+3rd) PRB installation at time steps of (a) 90 days, (b) 450 days, (c) 900 days and (d) 1800 days. Fig. 13 Percentage removal of Chloride (Cl-) in (a) Well 1, (b) Well 2, (c) Well 3, (d) Well 4, (e) Well 5, (f) Well 6, (g) Well 7, (h) Well 8, after installation of 1st PRB, 2nd PRB, 3rd PRB, (1st+2nd) PRB and (1st+2nd+3rd) PRB.

32

(b) 450 days

900

900

(a) 90 days

S2

C8

600

600

C8 C5

C6

C5 P

400

P

400

S2

C7

C3

C1 S1

C7

C4

200

200

C4

C6

C3

C1 S1

C2

-100

-100

C2

-100

0

300

600

900

1200

-100

1500

0

300

600

900

1200

1500

(d) 1800 days

900

900

(c) 900 days

600

S2

C8

C5

400

P

C6

S2

C5

C7

P

400

600

C8

C3

C1

C3

C1 S1

C2

C2

-100

-100

S1

C7

C4

200

200

C4

C6

-100

0

300

600

900

1200

1500

-100

0

300

600

900

1200

1500

(a) 90 days 900

900

(b) 450 days C8

S2

600

600

C8 C5

C6

C6

C7

400

C4

-100

0

200

C3

C1

300

600

900

1200

1500

-100

C3

C1

S1

C2

C2

-100

-100

200

C4 S1

C7

P

P

400

C5

S2

0

300

600

900

1200

1500

(d) 1800 days

900

900

(c) 900 days

C8

S2

600

600

C8 C5 C6

S2

C6

C7

C4

C1

200

200

C4 C3

C3

C1 S1

C2

C2

-100

-100

S1

C7

P

400

P

400

C5

-100

0

300

600

900

1200

1500

-100

0

300

600

900

1200

1500

(b) 450 days

900

900

(a) 90 days

C8

600

600

C8 S2

C5

C6

C5

C6

P

400

400

P

S2

C7

C4

200

200

C4 C3

C1 S1

C7

C3

C1 S1

C2

-100

-100

C2

-100

0

300

600

900

1200

1500

-100

0

300

600

900

1200

1500

(d) 1800 days

900

900

(c) 900 days

C8

600

600

C8 S2

C5

C6

S2

C7

C5

400

C6

P

400

P

C4 C3

C1 S1

200

200

C4

C7

C3

C1 S1

C2

-100

-100

C2

-100

0

300

600

900

1200

1500

-100

0

300

600

900

1200

1500

Percentage Removal

1st PRB

Percentage Removal

(1st+2nd) PRB

(1st+2nd+3rd) PRB

(b)

100

Well 1

80

Well 2

80

60

60

40

40

20

20

0

0 450

900

1350

1800

(c)

100

0

450

900

60

40

40

20

20

0

1800

Well 4

80

60

1350

(d)

100

Well 3

80

0 0

Percentage Removal

3rd PRB

(a)

100

0

450

900

1350

1800

(e)

100

0

450

900

60

40

40

20

20

0

1800

Well 6

80

60

1350

(f)

100

Well 5

80

0 0

Percentage Removal

2nd PRB

450

900

1350

1800

(g)

100

450

900

60

40

40

20

20

0

1800

Well 8

80

60

1350

(h)

100

Well 7

80

0

0 0

450

900

Time (days)

1350

1800

0

450

900

Time (days)

1350

1800

(a) 99.58 m

87.72 m

No Flow Boundary

C8

S2

C5

C6

C7

P

C4

C8

S1

C2

100.00 m

C1

88.00 m

No Flow Boundary

(b) 0m P C5

C6

C7

30 m

1st PRB

2nd PRB

3rd PRB

Constant Head Boundary

S-Contaminant Source

C-Concentration Well

P-Pumping Well

C8

C7

C6

C5

C4

C3

Concentration (mg/L)

(a) 3000

3000

Without PRB

2000

2000

1000

1000

1st PRB

0 0

450

900

1350

1800

0

450

(c)

Concentration (mg/L)

C1

(b)

0

3000

900

1350

1800

(d) 3000

2nd PRB

2000

2000

1000

1000

0

3rd PRB

0 0

450

900

1350

1800

0

450

(e)

Concentration (mg/L)

C2

3000

900

1350

1800

(f) 3000

(1st+2nd)PRB

2000

2000

1000

1000

0

(1st+2nd+3rd)PRB

0 0

450

900

Time (days)

1350

1800

0

450

900

Time (days)

1350

1800

3500 C1

C2

C3

C4

C5

C6

C7

C8

3000

Concentration(mg/L)

2500

2000

1500

1000

500

0 Without PRB

1st PRB

2nd PRB

3rd PRB

Without PRB Concentration (mg/L)

C8 Pumping Rate = P (a)

C7

C6

C5 C4 Pumping Rate = 5P (e)

3000

3000

2000

2000

2000

1000

1000

1000

0

0 450

900

1350 1800

0 0

450

1st PRB Concentration (mg/L)

(b)

900

1350

1800

0

3000

2000

2000

2000

1000

1000

1000

0

0 900

1350 1800

450

900

1350

1800

0

3000

2000

2000

2000

1000

1000

1000

0 900

1350 1800

1800

1350

1800

450

900

1350

1800

0

(h) 3000

3000

2000

2000

2000

1000

1000

1000

0 450 900 1350 1800 Time (days)

450 900 1350 Time (days)

1800

450

900 (l)

3000

0

900

0 0

(d)

0

1350

450

(k)

3000

450

1800

(g)

3000

0

1350

0 0

(c)

0

900 (j)

3000

450

450

(f)

3000

0

(1st+2nd) PRB Concentration (mg/L)

C2 C1 Pumping Rate = 10P (i)

3000

0

(1st+2nd+3rd) PRB Concentration (mg/L)

C3

0 0

450 900 1350 Time (days)

1800

0

C8

C7

C6

C5

5 Meter Depth Concentration (mg/L)

1st PRB (a)

10 Meter Depth Concentration (mg/L)

C2

3000

2000

2000

2000

1000

1000

1000

0

0 450

900

1350 1800

0 0

450

900

1350 1800

0

3000

3000

2000

2000

2000

1000

1000

1000

0 450

900

1350 1800

450

900

1350 1800

0

3000

2000

2000

2000

1000

1000

1000

0 900

1350 1800

450

900

1350 1800

0

(k) 3000

3000

2000

2000

2000

1000

1000

1000

0 450 900 1350 1800 Time (days)

450

900

1350 1800

(l)

3000

0

1350 1800

0 0

(j)

0

900 (i)

3000

450

450

(h)

3000

0

1350 1800

0 0

(g)

0

900 (f)

3000

0

450

(e)

0

C1

(1st+2nd+3rd) PRB (c)

3000

(d)

20 Meter Depth Concentration (mg/L)

C3

3000

0

30 Meter Depth Concentration (mg/L)

C4

(1st+2nd) PRB (b)

0 0

450 900 1350 1800 Time (days)

0

450 900 1350 1800 Time (days)

(a) 90 days 900

900

(b) 450 days C8

S2

600

600

C8 C5

C6

C5

C6

C4

200

200

C4 C3

C1 S1

C3

C1 S1

C2

C2

-100

-100

-100

0

300

600

900

1200

1500

-100

0

300

600

(c) 900 days

900

1200

1500

900

900

(d) 1800 days

S2

C8

600

600

C8 C5

C6

C5

C6

C4

200

200

C3

C1 C2

C3

C1

S1

C2

-100

-100

S1

C7

P

400

P

400

S2

C7

C4

-100

C7

P

400

P

400

S2

C7

0

300

600

900

1200

1500

-100

0

300

600

900

1200

1500

Highlights 

Multi-Reactive

Barriers

provide

highly

efficient

groundwater

remediation

performance. 

Location, orientation, and depth of installed barriers are important to determine PRB performance.



Barriers placed nearby to contaminant sources are more likely to remediate the aquifer.



Removal of the contaminant sequentially increased with the installation of multibarriers.



Continuous pumping increases the remediation rate of the barrier system in the vicinity.