in situ remediation (isr mt3dms local domain approach · in‐situ remediation (isr‐mt3dmstm)...

In‐Situ Remediation(ISR‐MT3DMSTM)

Local Domain Approach

1ISR‐MT3DMS Local Domain Approach

1 Local Domain

5 m

Global Domain

Global Domain

Sand

DNAPL

Well

Diffusion Into Clay

Silt or Clay

Forward Diffusion


Sand

Well

Back-Diffusion out of Clay

Back‐Diffusion

3

Silt or Clay

ISR‐MT3DMS Local Domain Approach

Factors Influencing Remediation Timeframe

4

THICK silt/clay:

‐ Sale et al., 2008‐Matrix Diffusion ToolKit(ESTCP, www.gsi‐net.com)

Influencing factors: ‐ Velocity‐ Thickness‐ Retardation‐ Diffusion rate‐ Transverse dispersion

‐ Length of clay lens‐ Biodegradation‐ Contact time


Introduction• Modeling diffusion‐dominated transport may require the addition of dozens to hundreds of layers to a model, depending on the thickness of silt/clay layers. This can have prohibitive costs, particularly for 3‐D models which already incorporate a large number of rows and columns.

• While there are analytical solutions for simulating diffusion in thicker layers of silt/clay, a numerical model is often needed for thinner silt/clay layers, or when complex degradation reactions occur.

ISR‐MT3DMS Local Domain Approach 5

Introduction• ISR‐MT3DMSTM offers the option to use local 1‐D domains to represent diffusion‐dominated transport. These 1‐D domains are outside the global model grid, and thus may result in significant cost and time savings for some sites.

• The local 1‐D domains incorporate the same reaction options that are available for the global model, so the effects of in‐situ bioremediation or ISCO, for example, may be simulated in silt/clay layers.

• Users also have the option to specify different horizontal and temporal discretization for the local 1‐D domains, relative to the global model.


Local domain(clay with limited extent, 50 layers)

Global domain

1 2 3

Each clay lens:10 to 100+ layers

100+ layers

Each clay lens:20 to 100+ layers

Water Table

Example Applications


0 1 2 3

Scale, in meters

0 1 2 3

Scale, in meters

Global domain

Local domain

8



0 1 2 3

Scale, in meters

Global domain

Local domain

9



Example #1: Ontario Site


Case Study #1 – Ontario Site

Treatment zone Waterloo Emitter

Groundwater elevation contour Other injection well

0 5 10

Scale, in meters

Sequenced Injections:1. Surfactant (NAPL)2. Hydrogen peroxide3. CaO24. Waterloo emitters


0 5 10

Scale, in meters

0 5 10 15 20 25 30 35 4090

92

94

96

98

100

Distance (m)

Section K-K'

Silty sand

Sand

Silt / Clay

Screened interval

Legend

WEST EAST

Sand


Model Grid

0 1 2 3

Scale, in meters

Minimum spacing = 4 inches(Waterloo Emitter diameter)

2‐D: 450 columns, 280 rows

Time step = 0.05 d

Phase I – 5 solutes(4‐hour run‐time)

0 5 10

Scale, in meters

Phase I – Waterloo Emitters horizontal influence


Phase I: Waterloo Emitters (t=3y)Case 1: PHC Koc = 5,000 mL/g

Case 2: PHC Koc = 50,000 mL/g

Electron Donors:

• GRO, DRO, Fe(II)

Electron Acceptors:

• DO, Fe(III)s

Reactions:

• Instantaneous or first‐order

• Reductive dissolution

Phase II model:

• Hydrogen peroxide• CaPO• GRO/DRO Conc.• Diffusion into silt


Case Study #1• Waterloo emitters used for passive oxygen injection, with approx. 1 meter spacing between wells

• Same wells also used for periodic active injections of oxygen‐releasing compound, chemical oxidant, or surfactant (depending on the event)

• Tight spacing of passive injection wells required high resolution grid discretization to evaluate zone of influence from passive injections

• Geology is interbedded sands with tight fine‐grained layers.

• Costs were prohibitive to develop a 3‐D model for evaluating vertical diffusion‐dominated transport in thick fine‐grained layers, given tight horizontal spacing and number of species in reactive transport model.


Local Domain Approach

16

Area of interest for modeling diffusion

Global Model Domain


Cross‐Section in Global Model (3 layers)

17

Sand Seam #1

SILT

Sand Seam #2

Source Area

Global model domain


Local Model Domains for Silt(1‐D Diffusion)

18

Sand Seam #1

SILT

Sand Seam #2

Area of InterestMultiple 1‐D vertical (Local) models

are linked to sand seamconcentrations in global model.

Silt layer is inactive to transport in global model.


Next Steps• Using the local domain approach substantially reduced the size of the global model domain.

• We are currently simulating the influence of active remediation on mass in the finer‐grained layers using the local domain approach.


Case Study #2: Florida Site


Approx. source zone extent

Site Characteristics

• Beach sand aquifer

• Continuous, thin clay layer across site

• Other discontinuous, thin silt/clay layers

• Multiple, thin suspended DNAPL layers in source zone

Extraction WellTransect

Injection WellTransect

Case Study – Florida Site


1

10

100

1,000

10,000

100,000

2002 2003 2004 2005 2006 2007 2008

TVOC Co

ncentration (ug/L)

Year

TCE MCL = 5 ug/L

Hydraulic isolation system started August 2002

Expected trendwithout back‐diffusion

Observed Trend

Modified from Parker et al., 2008

TVOC Trend After Source Containment


v = 130 ft/yαtv = 1.5 mm

v = 65 ft/yαtv = 1.5 mm

Distance (m)

Elevation (ft)

200 columns, 158 rows (layers)Minimum grid spacing: z = 1.25 cm, x = 0.5 m

Run‐time = 45 minutes for 85‐y simulation (t = 0.24 d)

Clay layer thickness = 0.2 m, foc = 0.5%

2‐inch thick TCE pool

2‐D Model Grid

23

Carey et al. (2015)


Distance (m)

Elevation (ft)

Clay

TCE pool: S=1100 mg/L, 5 m x 0.05 m

16 layersin clay

C=1,100 mg/L

t=35 y t=85 y0

TCESourceModel

Source Characteristics

24

DNAPL source removed at t=35 y.

Carey et al. (2015)


t = 0

Mclay = 136 kg

0 10 20 30 40 50 60 70 80 90 1000

5

10

0

0.005

0.1

1

10

100

0 10 20 30 40 50 60 70 80 90 1000

5

10

0 10 20 30 40 50 60 70 80 90 1000

5

10

Elevation (m

)Elevation (m

)Elevation (m

)

Distance (m)

TCEConcentration

(mg/L)

Mclay = TCE mass in clay assuming 20 m width.t = time since source removal.

t = 20 y

Mclay = 1.1 kg

t = 30 y

Mclay = 0.06 kg

30 years after source removal: 99.96% mass depletion in clay, avg. Cwell = 12 to 126 ug/L

Simulated TCE After Source Removal

25

Carey et al. (2015)


10987654321

5 m

1 Local Domain

5 m

(a) Local domain models with xLD = 0.5 m (b) Local domain models with xLD = 5 m

Conceptual illustration of local domains for two cases: (a) global and local domains have the same horizontal spacing; and (b) local domain has a larger horizontal spacing than the global domain grid.


vl

vu Hydrodynamic Dispersion (Dz = Dm + De)

Comparison of vertical mechanical dispersion (Dm) and effective diffusion coefficient (De) magnitudes in each grid cell of a 1‐D local domain. Vertical mechanical dispersion is shown to be significant at the top and bottom clay‐sand interfaces due to the use of a three‐dimensional dispersion tensor and horizontal velocity components at each clay‐sand interface. Application of a 1‐D diffusion model will result in underestimation of the mass flux between the transmissive zone and clay layer.

Elevation (m

)

De = Do

Dm = αtv v

Dz = 2.5x10‐5 m2/d

Dz = 1.7x10‐5 m2/d

Dz = 0.9x10‐5 m2/d


0.001

0.01

0.1

1

10

100

-35 -30 -25 -20 -15 -10 -5 0 5 10 15 20 25 30 35 40 45 50

Sim

ulat

ed W

ell C

once

ntra

tion

(mg/

L)

Time (y)

Simulated monitoring well concentrations at x=5, 25, and 100 m. Solid lines represent the global domain model, dashed lines represent the local domain model with local grid x=0.5 m, and dotted lines represent the local domain model with local grid x=5.0 m.

x=5 m

x=100 m

x=25 m


0

5

10

15

20

25

30

35

1 2 3 4 5 6

Remed

iatio

n Timeframe (y)

x = 5m x = 10m x = 25m x = 50m x = 75m x = 100m

Simulated remediation timeframe for three model cases: (a) no local domains are used; (b) 200 local domains are used with horizontal spacing of 0.5 m; and (c) 20 local domains are used with horizontal spacing of 5.0 m. Based on monitoring well with Lscreen=3 m.

200 Local domains 20 Local domainsNo Local domainsxLD = 0.5 m xLD = 5.0 m


0

5

10

15

20

25

30

35

0 20 40 60 80 100

Remed

iatio

n Timeframe (y)

Clay Layer Length (m)

(a) Remediation timeframe versus clay layer length (Lscreen=3 m)

(b) Remediation timeframe versus well screen length (x = 50 m)

0

5

10

15

20

25

30

35

40

0.1 1 10

Remed

iatio

n Timeframe (y)

Well Screen Length, Lscreen (m)


05

101520253035

0 5 10 15 20Remed

iatio

n Timeframe

(y)

Transverse Dispersivity, αtv (mm)

050

100150200250300350400

0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1Remed

iatio

n Timeframe

(y)

Clay Layer Thickness, Hclay (m)

foc = 1.5%R = 8.7

foc = 0.5%R = 3.5

(a) Remediation timeframe versus αtv.

(b) Remediation timeframe versus clay layer thickness (Hclay).

(c) Remediation timeframe versus v.

0

5

10

15

20

25

30

0 5 10 15 20Remed

iatio

n Timeframe (y)

Contact Time, tc (y)

(d) Remediation timeframe versus contact timebetween DNAPL and clay aquitard.

0

10

20

30

40

50

60

0.00 0.10 0.20 0.30 0.40Remed

iatio

n Timeframe

(y)

Groundwater Velocity, v (m/d)


1

10

100

1 2 3 4 5 6 7 8 9 10

Ratio

of M

axim

um to

Minim

umRe

med

ation Timeframe

Exhibit 14 – Comparison of relative sensitivity of remediation timeframe to various input parameters, based on the ratio of maximum to minimum timeframe for each set of modeled parameter adjustments. Hclay is the clay layer thickness, R is the retardation coefficient, v is groundwater velocity, Lclay is the length of the clay layer, Csol is solubility, is the tortuosity coefficient, and Lscreen is the monitoring well screen length. Based on clay layer length of 50 m and well screen length of 3 m unless except for Lclay and Lscreen parameter adjustments.

Hclay

R

v

Lclay αtv

Csol Lscreen

ClayHalf‐life Contact

Time


in situ remediation (isr mt3dms local domain approach · in‐situ remediation (isr‐mt3dmstm)...

Documents