z. .. -
‘ gages
, . V .-...,.i......5
. V _ . . £34.. ,3.
. V Ln. ..
r . , , . . ‘
.43.“. . , V .
.......:.m...,.V V V
,2 ‘ .
5% . .
. 34...?
V .?
7

3

Pariiukv 7:. 1.5 .:2....

in. i. 0.. ‘ . .HA
9....ny .52.... h a.

Mars. r
. x V t

{.43.}
...r18

.1... .Il...
5.1.53.
.a:..1:.. ..I.. .

Oahu-:5
hunk... 3 ..

an...) I
.3111...
1.1%.},

.2...
..!.171..

4.9.x

‘
4.2.9...
. I.\v:t

c. .
3.x:

.1. .
4.5.5.151.
.‘Ittvil
.31..»11
.2. A,

r. .9.

.- .JVMF

.2
v5

{1...

. I 3 . A . , I \

3. g. . R. : «:4

P159; . , V . V a .V. ‘ n V V .nmw V9.9.)
1...: :1... . , . ‘ ‘ V. :i

 

 

llllUllllUHIIHHUW:IUIHHHII‘JUIIWHHIIIIUW

193 01046 3473

This is to certify that the
dissertation entitled
Evaluation of Hydraulic Controls to Promote

Surfactant Dissolution of Trapped Residual NAPL

presented by

Lizette Rita Chevalier

has been accepted towards fulfillment
of the requirements for

Ph . D . degree in Civil Engineering

 

 

QM W%M

._ Major profes’ /

Date October, 1994

 

 

/ CL-Mbjor Professor

MS U is an Affirmative Action/Equal Opportunity Institution 0-12771

 

 

 

 

LIBRARY
Mlchlgan State
University

 

 

 

PLACE IN RETURN BOX to move this

Mutton mm
' To AVOID FINES return on or before due

date .

DATE DUE DATE DUE DATE DUE

    

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

EVALUATION OF HYDRAULIC CONTROLS TO PROMOTE SURFACTANT
DISSOLUTION OF TRAPPED RESIDUAL NAPL
By

Lizette Rita Chevalier

A DISSERTATION

Submitted to
Michigan State University
in partial fialfillment of the requirements
for the degree of

DOCTOR OF PHILOSOPHY

Department of Civil and Environmental Engineering

1 994

ABSTRACT

EVALUATION OF HYDRAULIC CONTROLS TO PROMOTE SURFACTANT
ENHANCED DISSOLUTION OF TRAPPED
RESIDUAL NAPL

by

Lizette Rita Chevalier

Hydraulic controls to deliver, recover and promote the efiicient use of aqueous
phase surfactants to enhance the dissolution of trapped residual non-aqueous phase liquids
near the water table were investigated through experiments and numerical modeling. Soil
column experiments were designed to simulate conditions near the capillary fiinge where a
non-aqueous phase liquid may become trapped as a discontinuous immobile phase. A
lumped mass transfer coefiicient for the dissolution of dodecane into a surfactant enhanced
aqueous phase was measured in a flow interruption study which determined that the mass
transfer is rate limited, but not prohibitively slow. Experiments were conducted to
measured the reduced relative permeability of the aqueous phase as a fimction of
saturation in a natural unconsolidated porous media where a trapped residual nonwetting
phase increases the resistance to aqueous flow. The data presented includes an
experiment where the trapped residual nonwetting phase was reduced through surfactant
enhanced dissolution in addition to measurements with a trapped residual intermediate
wetting phase present. An analytical technique was developed to measure the
concentration of dodecane in a surfactant using head space gas chromatography which
overcomes the effects of surfactant concentration on the degree of partitioning between
the gaseous and aqueous phase.

An implicit l-D transport code was developed to estimate the surfactant enhanced

dissolution of a uniform saturation of trapped residual non-aqueous phase. Graphs are

presented using dimensionless parameters to predict the time required for remediation and
to select a pumping rate promoting a high concentration of the contaminant while
minimizing the generation of waste water. A two-dimensional study was conducted in a
laboratory scale model aquifer using the United States Geological Survey model SUTRA,
a finite element code for saturated and unsaturated flow with contaminant transport. The
efficiency of the hydraulic systems was based on delivering the surfactant throughout the
contaminated region, ensuring sufficient contact time between the surfactant and the
contaminated region, and recovering contaminant laden surfactant enhanced ground water.
The results of the research show the promising potential and the need for further research
in surfactant enhanced pump-and-treat methods to remediate aquifers contaminated with

trapped residual non-aqueous phase liquids.

I dedicate this work, and the future work I do in the remediation of the earth and her
waters to the friends who died of cancer during the time I conducted this research ~ Joyce

Laderman, Helen Catherino and James Boggs.

iv

Acknowledgments

In respect, I would first like to thank my advisors Dr. David C. Wiggert and Dr.
Roger E. Wallace for their support, advice and criticism of my research. I would also like
to extend my appreciation to the other two members of my committee, Dr. Susan J.
Masten of the Department of Civil and Environmental Engineering and Dr. Raymond J.
Kunze of the Department of Crop and Soil Sciences for their advice and encouragement.
I would also like to thank Dr. Thomas Voice and Dr. Paul Loconto for their assistance.

I am grateful for the research funding provided by the US. Environmental
Protection Agency Ofice of Research and Development under a grant to the Great Lakes
and Mid-Atlantic Hazardous Substance Research Center. I would also like to thank the
College of Engineering at Michigan State University for the support received through the
Dean‘s Distinguished Fellowship. I also appreciate the support I received through a
General Electric Graduate Fellowship and an internship at General Motors NAO Research
and Development Center under the guidance of Dr. Abdul S. Abdul.

On the personal side, I gratefully thank my mother and proof-reader, Irene A.
Chevalier as well as my father, Leo N. Chevalier, for their encouragement and pride.

I appreciate the friendship and endless hours of conversation shared with many of
my colleagues, especially Herr Soeryantono, Reza Rakhshandehroo, Sashi Nair, Munjed
Maraqa and Jehng-Jyun (J .J .) Yao, all of which certainly enhanced the time spent at MSU.

My sincere thanks and love for the support and continued encouragement from the
Women in my life ~ Christine Reagan-Rosales, Ann Perrault, Sharon Howell, Sherry Hon,
Lucinda Lehmkuhle, Leah Cohen and most importantly, Lynette J. Holloway.

Table of Contents

List of Tables .................................................................................................................. ix
List of Figures ................................................................................................................. x
Nomenclature ............................................................................................................... xvi

Chapter 1. Introduction

1.1 Problem Statement ......................................................................................... 1
1.2 Objective and Scope of the Research .............................................................. 3
Chapter 2. Background
2.1 Irnmiscible Fluids at the Pore Scale ................................................................. 6
2.2 NAPL Entrapment as a Discontinuous Residual Phase .................................... 7
2.3 Hydraulic Conductivity and Permeability ........................................................ 8
2.4 Mobilization of Trapped Residual NAPL ...................................................... 10
2.5 Surfactants ................................................................................................... 13
2.6 Kinetics of Surfactant Enhanced Dissolution ................................................. 14
2.7 Surfactant Transport of Dissolved NAPL ..................................................... 15
Chapter 3. Materials and Methods
3.1 Apparatus .................................................................................................... 23
3.2 Preparation of Soil Column .......................................................................... 24
3.3 Establishing a Trapped Residual Irnmiscible Phase ........................................ 24
3.4 Hydraulic Conductivity Measurements ......................................................... 26
3.5 Equilibrium Solubility of NAPL in Surfactant ............................................... 27
3 .6 Investigation of the Kinetics of Surfactant Enhanced Dissolution
of Trapped Residual NAPL .................................................................... 27
3.7 Dye Sorption Measurements ........................................................................ 28
3.8 Error Analysis .............................................................................................. 28
3.9 Material ....................................................................................................... 29
Chapter IV. Column Experiments in Surfactant Enhanced Dissolution of Trapped
Residual Dodecane
4.1 Introduction ................................................................................................. 35
4.2 Objective ...................................................................................................... 37

vii

4.3 Background ................................................................................................. 37
4.4 Apparatus and Materials ............................................................................... 41
4.5 Methods ....................................................................................................... 41
4.6 Results and Discussion ................................................................................. 43
4.6.1 Analytical Method ............................................................................... 43
4.6.2 Equilibrium Solubility .......................................................................... 45
4.6.3 Kinetics of Dodecane Dissolution ........................................................ 46
4.6.4 Breakthrough Curve ............................................................................ 48
4.6.5 Long Term Continuous Pumping ......................................................... 50
4.7 Summary ...................................................................................................... 51
Chapter V. Aqueous Permeability Prior to and During Surfactant Enhanced Dissolution
of Trapped Residual NAPL
5.1 Introduction ................................................................................................. 59
5.2 Materials and Methods ................................................................................. 62
5.3 Results and Discussion ................................................................................. 64
5.3.1 Trapped Residual NAPL ...................................................................... 65
5.3.2 Reduced Permeability Of the Aqueous Phase Due to Trapped
Residual Phases ................................................................................... 68
5.3.3 Aqueous Phase Permeability During Surfactant Enhanced
Dissolution of NAPL ........................................................................... 69
5.3.4 Comparing Data in Discontinuous Region with Existing Fitting
Parameters .......................................................................................... 70
5.4 Summary ...................................................................................................... 72
Chapter VI. Numerical Modeling
6.1 Introduction ................................................................................................. 86
6.2 Objective ...................................................................................................... 89
6.3 One Dimensional Model Development .......................................................... 90
6.4 Simulation Studies Using l-D Implicit Transport Model ............................... 95
6.5 Verification of 2-D Numerical Code ............................................................. 98
6.6 Two Dimensional Model Simulations ......................................................... 101
6.7 Summary .................................................................................................... 108
Chapter VII. Dissertation Summary and Recommendations
7.1 Summary .................................................................................................... 147
7.2 Summary of Costs ...................................................................................... 152
7.2 Recommendations ...................................................................................... 152
Appendix A: Analytical Solution to the 1-D Transport Equation ................................. 155

Appendix B: ID Code for Surfactant Enhanced Dissolution of
an Irnmobile Residual NAPL ............................................................................ 156

Appendix C: Determination of Mass Transfer Coefficient:
Output fiom SCIENTIST Software .................................................................. 162

viii

Appendix D:Summary of Input Data used in SUTRA Model ........................................ 165
Appendix B: Materials and Suppliers ........................................................................... 169

List of References ........................................................................................................ 164

Table 2.1
Table 3.1
Table 4.1
Table 4.2
Table 5.1
Table 5.2
Table 5.3
Table 5.4

Table 6.1
Table 6.2

Table 6.3

Table 6.4

Table 7.1

Table C1.1

Table C1.2

Table D1,]

List of Tables

Overview of past research in NAPL entrapment and mobilization ........... 22
Fluid properties of dodecane ................................................................... 34
GC settings ............................................................................................. 58
Experimental set-up for breakthrough curves .......................................... 58
Capillary number vs. saturation ............................................................... 82
Soil and fluid properties .......................................................................... 83
Order of magnitude estimate for ganglia size ........................................... 84

Reported values of fitting parameters for white Ottawa sand used in
relative permeability-saturation and capillary pressure-saturation

relationships in a 2-phase system ............................................................. 85
Soil and fluid properties used in model simulations ............................... 144
Representative range of mass transfer coefficients ................................. 144

Error propagation for lumped mass transfer coefficient (k1),
saturatation of trapped residual NAPL (Sm). and

dispersivity (or) ..................................................................................... 145
Summary of well schemes for evaluation of hydraulic controls .............. 146

Summary of costs in 1986 dollars involved with in-situ leaching

(Roy F. Weston, Inc. 1988) .................................................................. 154
Output from Scientist Software used to determine mass transfer
coefficient, k1= 0.387 hrl. ................................................................... 162
Output from Scientist Software used to determine mass transfer
coefficient, k1= 0.161 hrl. ................................................................... 164
Data used in SUTRA numerical simulations .......................................... 165

Figure 1.1

Figure 2.1

Figure 2.2

Figure 2.3

Figure 2.4

Figure 2.5

Figure 2.6

Figure 2.7

Figure 2.8

Figure 3.1

Figure 3.2

Figure 3.3

Figure 3.4

List of Figures

Schematic of an unconfined aquifer with LNAPL
and DNAPL spills ..................................................................................... 5

Schematic of pore space characteristics ................................................... 17

Pressure difl‘erence at the interface of immiscible fluids in an idealized
pore throat or pore body ......................................................................... 18

Relative effect of the capillary number as a function of trapped
residual saturation at Bond numbers less than 0.00667
(adapted from Morrow and Songkran 1981) ........................................... 18

Relative permeability as a function of phase saturation ............................ 19

Flow passing through a permeable cylinder in a uniform flow field
relative to the flow passing through a cylinder when k, = 1 (adapted
from Wheatcrafi and Winterberg 1985) ................................................... 19

Relative correlation of the mobilization of residual oil and capillary number
for uniform glass beads (adapted from Morrow et al. 1988) .................... 20

Schematic of a surfactant micelle capturing and idealized NAPL ganglia. 20

Relative changes in physical properties in the neighborhood of the
critical micelle concentration (adapted fi'om Preston 1948) ..................... 21

1-D column ............................................................................................ 31

Schematic of apparatus used to conduct falling head hydraulic
conductivity measurements ..................................................................... 32

General schematic of apparatus used to measure hydraulic
conductivity ased on the head difi‘erence across the column ..................... 32

Grain size distribution of white Ottawa sand ........................................... 33

Figure 3.5

Figure 4.1

Figure 4.2

Figure 4.3

Figure 4.4

Figure 4.5

Figure 4.6

Figure 4.7

Figure 4.8

Figure 4.9

Figure 4.10

Figure 4.11
Figure 4.12

Figure 5.1

Figure 5.2

Figure 5.3

Adsorption isotherm for Red-O dye ........................................................ 33

GC response to dodecane concentration at varying surfactant
concentration illustrating the effects of surfactant concentration
on partitioning of dodecane .................................................................... 52

GC measurement of 4 ppm dodecane in varying surfactant
concentration with salt ............................................................................ 52

7 Four ppm undecane at varying surfactant concentrations ......................... 53

The influence of thermostat temperature and equilibrium time (temp/time)
on calibration curves ............................................................................... 53

Equilibrium solubility for dodecane for varying concentrations of

Witconol NP-150. .................................................................................. 54
Critical micelle concentration of Witconol NP-l 50 determined from

surface tension (Witco 1991) .................................................................. 54
Flow interruption study. ......................................................................... 55

Error associated with enhanced solubility measurement in flow
interruption study ................................................................................... 55

Continuous pumping of Witconol NP-l 50 followed by a water flush

and subsequent pulsed pumping .............................................................. 56
Breakthrough curves for 5% Mtconol NP-150 injected into regions

of trapped residual dodecane .................................................................. 56
Error associated with break through curves ............................................ 57
Compilation of breakthrough curves ....................................................... 57

Measured values of permeability-saturation for trapped residual air
and trapped residual NAPL ..................................................................... 75

Aqueous phase relative permeability as trapped residual NAPL is reduced
through surfactant enhanced dissolution .................................................. 76

Error associated with aqueous phase relative permeability measurements. 76

xii

Figure 5.4 Comparing data to Corey and van Genuchten equations for the
relative permeability-saturation relationship of the aqueous
phase with a residual aqueous phase saturation of 0.093 (A = 2.30,
n' = 1.55) ............................................................................................... 77

Figure 5.5 Comparing data to Corey and van Genuchten equations for the
relative permeability-saturation relationship of the aqueous
phase with a residual aqueous phase saturation of 0.093 (it = 2.32,
n' = 4.27) ............................................................................................... 78

Figure 5.6 Comparing data to Corey and van Genuchten equations for the
relative permeability-saturation relationship of the aqueous
phase with a residual aqueous phase saturation of 0.093 (X = 2.34,
n' = 6.99) ............................................................................................... 79

Figure 5.7 Comparing data to Corey and van Genuchten equations for the
relative permeability-saturation relationship of the aqueous
phase with a residual aqueous phase saturation of 0.093 (A = 2.30,
n '= 3.58) ................................................................................................ 80

Figure 5.8 Comparing data to Corey and van Genuchten equations for the
relative permeability-saturation relationship of the aqueous
phase with a residual aqueous phase saturation of 0.093 (A = 3.46,
n '= 7.42) ................................................................................................ 81

Figure 6.1 Implicit numerical solution compared to analytical solution
(Cr=0.5;L=15cm;ro=0.5;Da =1;Ax=0.5cm) ........................... 111

Figure 6.2 Maximum and minimum relative error over a range of velocities as
indicated by the range ofDa (Cr = 0.3; L = 15 cm; 0) = 0.6;
Ax = 1 cm). ........................................................................................ 111

Figure 6.3 Summary of study to optimize the Courant number over a
rangeofrovalues(L=15cm;Ar=1cm). ............................................ 112

Figure 6.4 Relative error over a range of Da numbers represented
in previous table (Cr = 0.3; (0 = 0.6; Q = 0.5 mllmin; .
L=15cm;Ax=1cm) .......................................................................... 112

Figure 6.5 Relative error over a range of Pa numbers (Cr = 0.3; a) = 0.6;
Q=0.5ml/min;L=15cm;Ax=lcm) ................................................ 113

Figure 6.6

Figure 6.7

Figure 6.8

Figure 6.9

Figure 6.10

Figure 6.11

Figure 6.12

Figure 6.13

Figure 6.14

Figure 6.15.

Figure 6.16
Figure 6.17

Figure 6.18

Figure 6.19

Figure 6.20

xiii

Fraction of trapped residual NAPL remaining in pores spaces
over the length of the column over time
(Da=10,Cr= 0.3;co=0.6;L=15cm;Ax=1cm) ............................ 113

Dimensionless analysis of surfactant enhanced dissolution
of trapped residual dodecane (Da = 1,Cr = 0.3; co = 0.6;
L=15cm;Ax=1cm) .......................................................................... 114

Dimensionless analysis of surfactant enhanced dissolution
of trapped residual dodecane (Da = 100,Cr = 0.3; 0) = 0.6;

L=15cm;Ax=lcm) .......................................................................... 114
Rate-limited dissolution of NAPL into surfactant enhanced

aqueous phase ...................................................................................... 115
Normalized effluent concentrations over a range of Da ......................... 115

Comparing eflluent concentrations when velocity is determined
from total porosity, effective porosity and variable effective
porosity ................................................................................................ 116

Experimental and implicit modeled effluent concentrations of
dodecane in a 5% nonionic surfactant enhanced aqueous phase ............. 116

Pores volumes needed to reach MCL over a range of

Damkohler numbers .............................................................................. 117
Model aquifer region. ........................................................................... 118
Elements in the saturated/unsaturated grid ............................................ 119
Nodes in the Saturated/Unsaturated grid ............................................... 120

Pressure-saturation relationship based on van Genuchten parameters 121

Schematic illustration of the water table and boundary conditions for
analytical solution of seepage through a dam (based on Wang and
Anderson 1979) .................................................................................... 121

Results of the comparison of SUTRA with analytical solution for
determining the surfact of the watertable through the aquifer ................ 122

Comparison of numerical and anlytical solution for transport of an
instantaneous point source .................................................................... 122

Figure 6.21

Figure 6.22

Figure 6.23

Figure 6.24

Figure 6.25
Figure 6.26
Figure 6.27
Figure 6.28
Figure 6.29
Figure 6.30

Figure 6.31

Figure 6.32
Figure 6.33
Figure 6.34
Figure 6.35
Figure 6.36
Figure 6.37
Figure 6.38

Figure 6.39

xiv

Geometry used in evaluation of well placement ..................................... 123

Concentration profiles as plume migrates through a region
of contamination where the relative permeability and effective
porosity are reduced due to trapped residual NAPL (Case 1a)

at t=1.67 hr, 3.4 hr, and 5 hr. ................................................................ 124
Concentration profiles as plume migrates through a homogeneous

medium (Case 1b) at t=1.67 hr, 3.4 hr, and 5 hr. ................................... 125
Summary of 5% concentration profile at 5 hours for Case 1a and

Case 1b ................................................................................................ 126
Velocity vector plot for Case 9 (Table 6.4) ........................................... 127
Velocity vector plot for Case 8 (Table 6.4) ........................................... 128
Velocity vector plot for Case 1a (Table 6.4) ......................................... 129
Velocity vector plot for Case 1b (Table 6.4) ......................................... 130
Concentration contour plot for Case 2 (Table 6.4) ................................ 131
Velocity vector plot for Case 2 (Table 6.4) ........................................... 132
Fraction of mass removed in each region after surfactant flushing

60 pore volumes ................................................................................... 133
Concentration contour plot for Case 3 (Table 6.4) ................................ 134
Velocity vector plot for Case 3 (Table 6.4) ........................................... 135
Concentration contour plot for Case 4 (Table 6.4) ................................ 136
Velocity vector plot for Case 4 (Table 6.4) ........................................... 137
Concentration contour plot for Case 5 (Table 6.4) ................................ 138
Velocity vector plot for Case 5 (Table 6.4) ........................................... 139
Concentration contour plot for Case 6 (Table 6.4) ................................ 140

Velocity vector plot for Case 6 (Table 6.4) ........................................... 141

Figure 6.40 Concentration contour plot for Case 7 (Table 6.4) ................................ 142

Figure 6.41 Velocity vector plot for Case 7 (Table 6.4) ........................................... 143

Nomenclature

A .................... cross sectional area of soil, L2

a ..................... cross sectional area of standpipe, L2

b ..................... width, L

C .................... aqueous phase solute concentration, ML’3

Cr ................... Courant number (dimensionless)

C s ................... aqueous phase equilibrium solute concentration, M];3
C“, ............... aqueous phase surfactant concentration, ML“3

c,- .................... distance from upper region of contaminated plume to well, L
D .................... hydrodynamic dispersion coefficient, L2T'l

D“ .................. coefficient of molecular difl’usion, LZT'1

Da .................. Damkohler number (dimensionless)

D, ................... fi'ee liquid difl‘usivity of the solute, L2T'l

d ..................... grain diameter, L

d, .................... distance from center of contaminated plume to well, L
Fa, ................. fraction of flow reduced (dimensionless)

F g ................... gravity forces, MLT"2

F r ................... interfacial tension forces, MLT'2

Fv ................... viscous forces, MLT'2

g ..................... gravitational constant, LT'2

Hi ................... hydraulic head, L

i ..................... hydraulic gradient (dimensionless)

J ..................... flux of mass across interface, ML'ZT'1

K .................... hydraulic conductivity, LT'l

k ..................... permeability, L2

k, .................... lumped mass transfer coefficient, LT‘l

I ..................... characteristic length, L

L .................... length of column, L

Ly ................... vertical length of ganglia, L

L}, .................. horizontal length of ganglia, L

m .................... fitting parameter

M ................... mass, M

MC .................. column mass, M

Md .................. dry mass of packed soil column, M
MW ................. mass of water in soil column, M
we ................. mobility ratio" (dimensionless)

n ..................... porosity (dimensionless)

n' .................... fitting parameter

Nb .................. bond number (dimensionless)

Nc .................. capillary number (dimensionless)

P .................... pressure, ML'lT'2

PC ................... capillary pressure, ML'lT'2

Pe .................. Péchlet number (dimensionless)
PV .................. pore volume, L3

PV" ................ aqueous phase pore volume after trapped residual NAPL, L3
PV; ................. initial pore volume, L3

Q .................... flow, L3 T"1

Q; ................... scaled source/sink (dimensionless)
Qa .................. flux through contaminated region
R .................... grain radius, L

82 .................. uncertainty interval (residual error)
Re .................. Reynolds number (dimensionless)
S ..................... saturation (dimensionless)

s», ................. standard error of the y estimate

1 ..................... time, T

1w,” ............ Student’s t

t" .................... dimensionless time, pore volume

v ..................... average pore velocity, LT“l

Vc ................... volume of column, L3

Va,p ................ volume in column caps, L3

w .................... width of contaminated region, L
x} ................... scaled x—distance fi'om centerline of contaminated region (dimensionless)
x,- .................... Cartesian coordinates, L

y',- ................... scales y-distance from top of contaminated region (dimensionless)

y ..................... Cartesian coordinates, L

xviii

z ..................... elevation head, L

Greek

p .................... density, ML'3

A .................... difference

a .................... dispersivity, L

a' .................... fitting parameter

A. .................... fitting parameter

(a .................... implicit weighting parameter

a .................... interfacial tension, ML‘ZT'2

e ..................... relative error

t ..................... stress, ML'IT'2

6 ..................... thickness of stagnant film layer, L
if .................... viscosity, ML‘IT'l

Subscripts

a ..................... aqueous phase

b ..................... bulk

e ..................... effective

g ..................... gaseous phase

ma .................. maximum aqueous phase

It ..................... NAPL

mv .................. nonwetting

p ..................... phase

r ..................... relative

ra ................... irreducible aqueous phase

rg ................... trapped residual gaseous (air) phase
rn ................... trapped residual NAPL (oil) phase
s ..................... saturated

w .................... wetting

Chapter 1
Introduction

1.1 Problem Statement

Surface spills, tank leaks and improper disposal practices may result in aquifer
contamination by a pollutant immiscible with the groundwater, such as gasoline. As the
volume of nonaqueous phase liquid (N APL) advances under the influence of gravity
through the porous matrix, a residual volume will remain in the pores, trapped by
capillary forces. In the upper region of an unconfined aquifer, the NAPL will migrate
through pores initially saturated with a continuous aqueous and gaseous phase, referred
to as the vadose zone (Fig. 1.1). In this region, as much as 9% of a pore space will
become occupied with residual NAPL (Conrad et al. 1987) as the plume moves
downward to the capillary fringe, where the pores are largely saturated by the aqueous
phase. Once the plume reaches this region of the aquifer, NAPL that is more dense than
water (DNAPL) will continue to migrate through the aquifer trapping discontinuous
residual in the pores spaces unless an impermeable boundary impedes migration, causing
the DNAPL to pool (Schwille 1967). NAPL less dense than water (LNAPL) will pool as
a lens above the capillary fiinge after migrating through the vadose zone. As the water
table fluctuates in response to local pumping or seasonal recharge and discharge, the lens
of LNAPL at the capillary fiinge can become smeared over a larger region as smaller
disconnected pockets of trapped residual NAPL (Hunt et a1. 1988). Regardless of
whether the transport of NAPL results in the formation of a lens or a residual, NAPL
components and chemicals will partition into the aqueous and gaseous phases present in

the aquifer. On average, the maximum permissible level of a NAPL contaminant

acceptable in water is defined by the US. Environmental Protection Agency as more than
2 orders of magnitude less than the solubility of typical NAPL contaminants (Miller et al.
1990). As a result, NAPL dissolution into the groundwater can contaminate an aquifer
for decades, producing a dilute waste stream of massive volume (Hunt et al. 1988).

The field application of in-situ remediation technologies for the treatment of
NAPL has not achieved the high efficiency suggested by ID column experiments
associated with treatability studies. This is due in part to the fact that l-D columns
impose an artificial boundary that controls fluid movement between injection and
recovery points. In the field however flow is free to by—pass the region contaminated
with NAPL due to the resistance caused by the immiscible oleic phase within the soil
pores. As a result, the efliciency of a treatment process for NAPL that relies on transport
through the aqueous phase may be dramatically decreased in the 3-D in-situ environment.

Compared to a lens of continuous NAPL, the resistance to aqueous flow is
reduced for trapped discontinuous residual NAPL below the capillary fiinge. In this
region, trapped residual NAPL occupies 10-50% of the pore space (Melrose and
Brandner 1974). As a result, the resistance to the flow of the aqueous phase may still be
significant enough to cause preferential flow around the contaminated region. The
remediation of trapped reSidual NAPL is more arduous, as the NAPL is trapped by
capillary forces stronger than the combined opposing viscous and buoyant forces. At the
capillary fringe, the presence of a trapped residual gaseous phase may further accentuate
the by-passing problem due to the reduced aqueous permeability in areas contaminated
with trapped residual NAPL.

Current remediation technologies have not identified practical methods for
removing trapped residual NAPL in the saturated zone (Palmer and Fish 1992; Keely
1989). The proposed study will consider chemically enhancing the aqueous phase with
surfactants to promote the dissolution of NAPL into the aqueous phase. For laminar flow

conditions typical of aquifers, the dissolution of the trapped residual NAPL may be a rate-

limited process resulting in less than predicted levels of NAPL contamination and clean-

up efficiency (Hunt et al. 1988).

1.2 Objective and Scope of the Research

The overall objective of this research is to evaluate hydraulic controls to promote
the flow and residence time of surfactant enhanced water through regions contaminated
with residual NAPL. The research will incorporate l-D column experiments, the
development and use of a 1-D implicit transport model and the use of a 2-D numerical
model with emphasis on 1) problems associated with the developing an efficient in-situ
delivery and recovery design; and 2) trapped residual NAPL at the capillary fiinge. To
reach these objectives, the research is conducted in six tasks, each of which define a

specific objective of the research.

TASK 1 Develop analytical methods for measuring NAPL concentration in a
surfactant enhanced aqueous phase. 7

TASK 2 Measure the relative permeability-saturation relationship for the aqueous
phase in the region where NAPL is no longer a continuum.

TASK 3 Measure rate-limited kinetics of surfactant enhanced dissolution of NAPL
using a flow interruption technique.

TASK 4 Measure the breakthrough of a surfactant enhanced aqueous phase injected
into a region of trapped residual NAPL.

TASK 5 Develop a 1-D implicit transport code to model the dissolution of trapped
residual NAPL in column experiments in order to view the removal of NAPL
as a fiinction of time and space and evaluate pumping rates.

TASK 6 Use the United States Geological Survey model for Saturated-Unsaturated
Transport (U SGS SUTRA) to evaluate the efficient use of well locations as a

hydraulic control in conjunction with the pumping rates optimized in Task 5

to promote the delivery/recovery of the surfactant enhanced aqueous phase

into a region contaminated with trapped residual NAPL.

In Task 1, an analysis method will be selected and developed after evaluating .
liquid-liquid extraction methods, liquid-solid phase extraction methods and head space
analysis methods. In Task 2, a trapped residual NAPL phase and/or a trapped residual air
phase will be established in 1-D soil columns to simulate the region below the capillary
fringe. Falling head permeability measurements will be used to compared the reduced
permeability over a range of trapped residual NAPL saturation. In addition, a long term
dissolution study will be conducted to measure increased aqueous phase permeability as
the immiscible trapped NAPL is reduced through dissolution. In Task 3, a lumped mass
transfer coefficient will be measured to predict the rate-limited dissolution of NAPL into
a surfactant enhanced aqueous phase. A flow interruption technique common to
measuring rate-limited processes such as thin film difliision or sorption will be adapted.
In Task 4, a surfactant enhanced aqueous phase will be injected into a column with a
uniform distribution of trapped residual NAPL in order to evaluated the dispersion of
surfactant within a region contaminated with trapped residual NAPL. Tasks 2-4 will
provide design parameters for the numerical modeling and remediation design developed
in Tasks 5-6. In Task 5, the lumped mass transfer coefficient measured in Task 3 will be
used to account for the dissolution of NAPL into a surfactant enhanced aqueous phase in
a 1-D finite difference code which will emphasize modeling the removal of NAPL as a
function of time and space. This differs from the approach of evaluating treatment
techniques of predicting the rate of recovery fiom the effluent only, which essentially
treats the soil column as a batch reactor. In Task 6, a region of contamination will be
viewed as region of reduced aqueous permeability due to the trapped residual NAPL.
Hydraulic controls will be evaluated and selected based on: 1) establishing a zone of

influence which promotes flow into the contaminated region; 2) recovery of contaminant

laden fluid; and 3) promoting the contact time needed to reach equilibrium or near
equilibrium dissolution of NAPL into the surfactant enhanced aqueous phase, assuming

that the flow paths are independent of chemical treatment.

 

ground surface

......
......
.....
..........

.......
......
.........
............

.......

............................................

 

   

capillary
fringe

 

 

water table v

DNAPL

 

 

 

Figure 1.1. Schematic of an unconfined aquifer with LNAPL and DNAPL spills.

Chapter 2
Background

2.1 Immiscible Fluids at the Pore Scale

As most soils are hydrophobic, the surface of the soil grains maintains a
continuous aqueous phase throughout the porous matrix. Within the pore space, the
saturation of the aqueous phase will vary between pores saturated with the aqueous phase
to soil pores where only a thin film of water may be present on the solid surface of the
soil grain. A fluid immiscible with the aqueous phase, such as a gaseous air phase or a
NAPL (e.g., gasoline), may also be present, but not necessarily as a continuous phase in
all regions. In these hydrophilic soils, the aqueous-NAPL-gaseous (water-oil-air) phases
are oriented in a soil pore as shown in Figure 2.1 (Corey 1984). The aqueous phase that
preferentially wets the soil is referred to as the wetting fluid. In a two phase system, the
second fluid is referred to as the nonwetting fluid. In three-phases, the phase between the
wetting and nonwetting fluid is referred to as the intermediate wetting fluid. Figure 2.1
illustrates this wetting pattern in two phase systems with NAPL-gaseous phases,
aqueous-gaseous phases and aqueous-NAPL phases. The order of the fluids with respect
to the solid grain surface plays an important role in understanding the transport of the
various phases through the porous matrix. ‘

The wetting fluid is held in the soil pores by capillary pressure. The capillary
pressure, PC, is a measure of the difference in pressure across the interface of the wetting
and non-wetting fluids in the soil pores, P,,w - PW. It is a function of the interfacial

tension between the fluids, 0, and the radius of curvature of the interface, r (Fig. 2.2).

P=P —P =— (2.1)

In a NAPL spill in a hydrophilic sandy aquifer, typically, the aqueous phase is the
wetting fluid and NAPL the nonwetting fluid. In aqueous-gaseous phase systems, the
bubbling point pressure, or air entry pressure, has been defined as the pressure needed to
force the gaseous phase into a capillary tube while draining the water out. A similar
definition applies to NAPL displacing the aqueous phase. The NAPL entry pressure is
the pressure needed to displace the aqueous phase held in the soil pores by the capillary
pressure. Since capillary pressure is inversely proportional to the radius of curvature, a
higher NAPL entry pressure is needed to displace the aqueous phase held in the smaller
pores and pore throats. As a result, NAPL penetrating a region where the aqueous phase
saturates the pores will displace the aqueous phase from the larger pores first. However,
when the aqueous phase is reintroduced into the pores, the stronger capillary forces in the

smaller pores and pore throats will pull the aqueous phase into these pores spaces first.

2.2 NAPL Entrapment as a Discontinuous Residual Phase

The petroleum industry has studied the mechanisms of residual oil entrapment and
recovery for secondary and tertiary recovery. A brief overview of this work is presented
in this section. The distribution and trapping of trapped residual NAPL, or ganglia, is
influenced by the geometry of the pore spaces, fluid and soil interfacial properties, the
interfacial tension between the fluids, pressure gradients and gravitational forces (Morrow
and Songkran 1981). These factors can be evaluated in two dimensionless numbers
which incorporate viscous (F v), buoyant or gravity forces (Fg) and interfacial tension

([F'I') forces (Ft).

Fg = mg = Apl3g (2-2)

Fv = 1A = p[$)A Ep(;)lz =pvl (2.3)

Ft = 01 (2.4)

where m is mass, g the gravitational constant, Ap the density difference, I stress, A cross
sectional area, it viscosity, v velocity, 1 a characteristic length, and o interfacial tension.
The capillary number, No, reflects the ratio of viscous forces to capillary or surface
tension forces , whereas the bond number is the ratio of buoyant or gravity forces to

capillary forces (Morrow and Songkran 1981; Larson et al. 1981).

Nc=L‘_‘-_-kw_Ae (2.5)
a la
2
NIL—Jill}: (2.6)
a

where kw is the permeability of the wetting phase and R the radius of a grain of soil.
Morrow and Songkran (1981) found that for Nb less than 0.00667, mobilization
and entrappment are more strongly influenced by changes in the ratio of the capillary to
viscous forces as opposed to the relationship between the buoyancy and capillary forces
(Fig. 2.3). Table 2.1 reviews past research in trapped residual NAPL which indicates that
a range of trapped residual NAPL saturation (Sm) can exists in sands. A similar review is

presented in Mercer and Cohen (1990).

2.3 Hydraulic Conductivity and Permeability

Hydraulic conductivity, K, is a measure of the resistance to fluid flow through

porous media. It is a function of both the fluid and soil properties:

K: has

(2.7)
p

where It is the permeability of the soil, p is the density of the fluid, u is the viscosity of
the fluid and g is acceleration due to gravity. In a two- or three-phase system,
permeability is also considered to be a fiinction of the fluid phase saturations (Fig. 2.4). In

the absence of the nonwetting fluid, the relative permeability of the wetting fluid is 1.
Once the nonwetting fluid is introduced, the permeability of the wetting fluid decreases as
the saturation of nonwetting fluid in the pores increases. Conversely, if the wetting fluid is
reintroduced into the pore space the permeability of the wetting fluid will increase as the
saturation of the nonwetting fluid decreases. At low hydraulic gradients common to
ground water systems, water reintroduced into the soil pores cannot completely remove
the non-wetting fluid (Wilson et al. 1990). Instead, the non-wetting fluid becomes
immobile, discontinuous and trapped by capillary pressures stronger than the combined
viscous and buoyant forces. Within these pores, the maximum saturation of water (Sm)
is limited by the trapped residual saturation of the non-wetting fluid, which in turn limits
the permeability of water. As discussed previously, the aqueous phase is a continuous
fluid within the porous matrix. The residual saturation of the aqueous phase is defined as
the irreducible aqueous phase where the aqueous phase is at a minimum (Sm). The
trapped residual NAPL phase is defined as Sm in Figure 2.4.

If permeability in a contaminated region is reduced due to the oil present in the
soil pore, streamlines will deflect at least partially around the region, resulting in a smaller
flux of water through the contaminated region. Wheatcraft and Winterberg (1985)
showed that the fiaction of flow that would pass through a cylinder of reduced
permeability relative to the amount that would pass through the cylinder if the entire

region was homogeneous in a uniform flow field can be calculated from:
F“ = 2k,/(1+ k,) (2.8)

where k, is the permeability of the cylinder relative to the permeability of the surrounding
medium (Fig. 2.5). Consider a circular region where the permeability is reduced (i.e.,
regions where k, = 0.6, 0.3, 0.1). A comparison of the fraction of flow, Fm, through
these circular regions of reduced permeability with flow passing through a comparable
cross-section where k, = 1 resulted in values of Fa, equal to 0.75, 0.46, and 0.18 for the

10

respective three regions of reduced permeability. These preliminary results support the
need to evaluate the reduced permeability when designing an in-sr’lu ground water
technology targeting immiscible contaminants.

The use of hydraulic controls to promote flow paths through regions of reduced
conductivity in the field are not well understood. Analytical and numerical solutions to
groundwater flow problems have traditionally focused on the flow paths created by
hydraulic controls in homogeneous aquifers (Bear 1979; White 1979). These studies can
provide the basis for evaluating the initial placement of injection/recovery wells and
pumping rates. The results will be different, however, once the heterogeneous region

resulting fi'om the reduced permeability of pores with trapped residual NAPL is included.

2.4 Mobilization of Trapped Residual NAPL

Past research in both petroleum engineering and environmental remediation
engineering has focused on removing trapped residual NAPL through mobilization. With
mobilization, the trapping of residual NAPL is controlled by capillary forces stronger than
the combined viscous and buoyant forces present. Past remediation research has
considered reducing No by either hydraulically increasing the gradient (Wilson and
Conrad 1984) or chemically enhancing the groundwater to reduce the interfacial tension
(Texas Research Institute 1979; Abdul et al. 1992; Ellis et al. 1984; Nash and Traver
1984).

Wilson and Conrad (1984) considered increasing the hydraulic gradient in both
horizontal and vertical sweeps to mobilize residual hydrocarbons in the saturated zone.
They concluded that for k > 100 cm2 (coarse sand and fine gravel), residual blobs could
be mobilized in both horizontal and vertical sweeps. In addition , for k > 1 cm2 (sandy silt
or fine sand), residual blobs could also be mobilized in a vertical sweep, but not in
horizontal sweeps. For k < 0.1 cm2 (glacial till, fine silt, clay), sufficiently high gradients

cannot be achieved, vertically or horizontally, to mobilize residual oil in the saturated

11

region. As an applied technology, this method is limited by the practical constraints of
establishing the higher hydraulic gradients in the field (Wilson et. al 1990).

The capillary number can also be increased by lowering the IFT between the
NAPL and water sufficiently to allow the capillary force to overcome the combined
viscous and buoyant forces so that oil can be released from the pore space. Petroleum
engineers have measured the reduction in the residual trapped NAPL (Sn/8*m) as a
function of the capillary number (Fig. 2.6) . In studies of sandstone and glass bead packs,
residual oil was completely removed at Nc > 10'2 (Chatzis and Morrow 1981,1984;
Morrow and Chatzis 1982; Morrow et al. 1988). The critical Nc is defined as the point
where mobilization of the residual non-wetting fluid begins. Wilson et al. (1990)
measured the critical No in Sevilleta sand at Nc = 104, which falls between the critical Nc
for sandstone and glass beads (10'5 and 103, respectively). Surfactant enhancement of
the aqueous phase can alter the interfacial tension by 4 orders of magnitude (Fountain et
al. 1991), which may reduce the Ne sufficiently to cause mobilization. By enhancing the
aqueous phase with surfactants to lower the interfacial tension, the residual can be
' mobilized. Ng et al. (1978) showed that mobilized residual oil will migrate into adjacent
pores and coalesce with an existing ganglia. As released residual continues to coalesce in
this fashion, the permeability of water will begin to decrease as the oil saturation
increases. Eventually, the decreased permeability will significantly reduce the flux of
surfactant into the pores (Texas Research Institute, 1979). This loss in efficiency results
in an increased cost for the technology. In the case of residual DNAPL trapped within
and below the capillary fiinge, the released contaminant threatens to migrate deeper into
uncontaminated regions of the aquifer.

The problems associated with IFT surfactants and delivery are suggested in a well
documented pilot scale study by the U. S. Environmental Protection Agency at the Volk
Air National Guard Base, Camp Douglas, Wisconsin. This in-sr'tu remediation targeted

the use of surfactants to remove NAPL in the vadose zone. Prior column studies

12

suggested removal efficiencies as high as 98% after flushing with 12 pore volumes of
surfactant (Nash and Traver 1984). However, the in-situ pilot scale was not measurably
effective in removing 'the contamination after 14 pore volumes. The final report could not
cite conclusive reasons for the pilot scale failure. It does cite a plausible explanation,
namely that surfactant may have flowed in preferential pathways created in regions
containing smaller fractions of the more highly contaminated fine soil (Nash and Traver,
1985). In addition, if the soil contained a high level of fines, the surfactant may have
caused a dispersion of colloidal size particle which could lead to clogging of the soil
pores and a reduction in flow (Abdul et al. 1990). However, an alternative explanation
can be found if one views the reduced aqueous flow within pore spaces resulting from the
coalescing of released residual NAPL. The increased saturation of NAPL collecting in
the pores spaces will create a blockage, resulting in reduced permeability. Unrestricted
by a boundary (i.e. column walls) the flow is free to flow through areas of higher
conductivity around the contaminated regions.

Petroleum engineers have combined polymers with surfactants to aid in mobility
control of the released NAPL by increasing viscosity while mobilizing residual oil (Sale
and Pitts 1989; Smith 1966). The significance of increased viscosity can be explained by
examining the mobility ratio, MR.

_ mobility of displacing fluid _ km ,u" (2 9)
mobility of displaced fluid ‘ km p, '

 

where Iran, is the effective permeability of the aqueous phase at the residual NAPL
saturation, [rum is the effective permeability of the NAPL at residual aqueous phase
saturation, ua is the viscosity of the aqueous phase and w: the viscosity of the NAPL.

As the change in interfacial tension mobilizes residual oil, the decreased saturation
of the residual oil will result in an increase in the permeability of water. As a result, the

mobility ratio will increase. This increase may result in fingering of the displacing fluid

13

past the displaced fluid, minimizing the contact of the surfactant with the contaminated
region. The addition of polymers to increase the viscosity of the aqueous phase is
intended to counterbalance the increase in aqueous permeability. A surfactant-polymer
pilot scale study was conducted to remove residual DNAPL at a former wood testing site
in Wyoming with reasonable success (Sale et al. 1989). In this system, an impermeable
lower boundary provided a barrier to released DNAPL migrating into uncontaminated

regions of the aquifer.

2.5 Surfactants

Surfactants, or surface active agents, have a characteristic amphipathic molecular
structure consisting of a lyophobic structural group that has very little attraction for the
solvent together with a lyophilic structural group which has a strong attraction for the
solvent (Rosen 1989). In water soluble surfactants intended to enhance the dissolution of
NAPL, the solvent is water and the solute is NAPL. In aqueous system, the lyophobic
and lyophilic structural groups are ofien referred to as hydrophobic and hydrophilic.

At low concentrations in water, surfactants begin to orient with the hydrophilic
ends toward and the hydrophobic end repelled away from the aqueous phase. Eventually,
the critical micelle concentration (CMC) will be reached. At this concentration, colloidal
sized clusters called micelles begin to form which can solubilize nonaqueous fluids into
the aqueous phase (Fig. 2.7) At the CMC, breaks in the trend of other measurable
physical properties including surface tension, detergency, osmotic pressure, equivalent
conductivity and interfacial tension also occur (Fig. 2.8). The micelles capture the NAPL
by surrounding it with the hydrophobic ends. Once trapped in the micelle, the NAPL
moves in the same phase as the water. As a result, surfactants make it possible to
solubilize normally insoluble substances. Solubilization can be distinguished fi'om
emulsification, which is the dispersion of one phase into another, with each phase

remaining distinct. As the amount of NAPL is increased in the interior hydrophobic core,

14

a micelle will become more and more asymmetric, eventually becoming lamellar in shape.
In theory, this process can be reversed by adding water to decrease the concentration of
surfactant.

Interfacial tension is a measure of the minimum amount of work required to create
an interface. At the interface, molecules have a higher potential energy than interior
molecules due to the fact that molecules interact more strongly with molecules in the
interior. Work is required to bring the molecules to the surface. When water is enhanced
with surfactant, the hydrophobic group will increase the fi'ee energy of the system. This
means that less work is required to bring a surfactant molecule to the surface than a water
molecule. As a result, surfactants concentrate at the interface, with the hydrophilic tails
oriented inward and the hydrophobic tails at the interface (Rosen 1989).

The use of surfactants which increase the solubility of NAPL into the aqueous
phase addresses some of the problems associated with IFT surfactants, yet there is still a
threat of spreading the contamination to new regions if the flow carrying contaminant
bearing micelles is not recovered. Consequently, the need to establish hydraulic control is
essential for recovery as well as delivery of the surfactant. In addition, recovering the
surfactant for recycling can lower the costs of in-situ surfactant flushing (Clarke et al.
1992).

2.6 Kinetics of Surfactant Enhanced Dissolution

Ifthe flow rates during remediation are too rapid to allow the surfactant enhanced
water sufficient contact time with residual NAPL for the surfactant to achieve the
maximum solubility of NAPL, the technology will produce large volumes of mildly
contaminated water. As a result, the expense involved with both the surfactant and the
treatment of the waste water may render the technology prohibitive from an economical
viewpoint. Intermittent (pulsed) pumping has been suggested to address the inefficiency

resulting from non-equilibrium (Keely 1989). In this pump and treat method, the cycling

15

of injection/extraction well pumping schedules is based on promoting the residence time
needed for water to come into localized equilibrium with trapped residual NAPL.
Preliminary studies were conducted by Pennell et a1. (1992) to measure the
solubility of dodecane in surfactant as a function of pore water velocity and the duration
of flow interruptions. They concluded that pulsed (intermittent) pumping is more efficient
than continual pumping. In contrast, Ang and Abdul (1991) conducted two 1-D column
experiments comparing intermittent vs. continuous pumping which showed no significant
difl‘erence in surfactant removal efficiency for removing residual automatic transmission
fluid from sandy soil. A numerical simulation study by Borden and Kao (1992) of
intermittent pumping concluded that the greater amount of hydrocarbon recovered per
volume water may increase the time required to meet a specific remediation standard.

These different conclusion emphasize the need to conduct site specific studies.

2.7 Surfactant Transport of Solubilized NAPL

When a continuous tracer is introduced into the aqueous phase of the porous
media, it will be advected by the groundwater flow. In addition, it will also spread out
laterally and longitudinally, occupying an ever increasing portion of the flow domain.
This spreading phenomenon is called hydrodynamic dispersion (dispersion, miscible
displacement) in groundwater (Bear 1979). The one-dimensional form of the advection-

dispersion equation for a non reactive dissolved constituent is

2
at 3:: fix

 

for saturated, homogeneous, isotropic materials under steady-state, uniform flow (Freeze
and Cherry 1979). In this equation, C is the concentration of the solute in the aqueous
phase, D is the coefiicient of hydrodynamic dispersion, v is the average linear velocity, 1 is

time and x is the spatial coordinate.

16

This equation may be written in the non-dimensional form (Brusseau and Rao

1989)

aC'_ié’2C’_§C°
at. Pe 6X2 é’X

 

 

(2.11)

by introducing the following dimensionless parameters:

(3.ng Pe=%
° (2.12)
. vi

t = I X =%
In Eqn. 6.11-12, C * is the dimensionless concentration, C0 equilibrium concentration, Pe
the Péclet number, L the length of the column, t* dimensionless time (pore volume), and
X dimensionless distance.

The slope of the breakthrough curve, which is a measure of the magnitude of the
dispersion, is a fiinction of Fe. The coefficient of hydrodynamic dispersion may be
determined fi'om the slope of the front part of the breakthrough curve. It can be

expressed in terms of two components:

D: m7+D' (2.13)

where dispersivity, a, is a characteristic property of the porous media, and the coefficient
of molecular difi‘usion, D', is a characteristic of the solute . At high velocity, dispersion is
dominated by mechanical mixing relative to the effects of molecular diffusion. At low
velocity, dispersion due to molecular diffusion dominates (Freeze and Cherry 1979).
When an aqueous surfactant is continuously injected into a region of the
groundwater system to enhance the aqueous solubility of an immiscible residual NAPL, it
will exhibit the same behavior as that of a tracer, with dispersion influenced by the
properties of the fluids and the porous media. A breakthrough curve measuring the

concentration of NAPL in the surfactant will provide a measure of the zone of influence

17

created by the transport of NAPL solubilized into the surfactant enhanced aqueous phase.
However, the chemical processes of surfactant solubilization in the porous matrix are also
influenced by surfactant sorption to the soil, which may also exhibit nonequilibrium.
Theoretically, surfactant sorption without nonequilibrium will shift the breakthrough
curve to the left. In contrast, mass transfer limitations due to dead-end pore spaces and
increases in local flow velocities as a result of trapped residual NAPL restricting flow
paths can shift the breakthrough curve to the right. As a result, a 1-D column study to
measure the breakthrough curve of surfactant injected into soil with trapped residual
NAPL integrates the complexity of different chemical and flow processes of dissolution,

sorption and variability in ganglia size and shape.

 

 

        

 

 

 

Figure 2.1. Schematic of pore space characteristics.

18

 

 

 

 

 

Figure 2.2 Pressure difference at the interface of immiscible fluids in an idealized
pore throat or pore body.

 

 

 

16 a—
14 —~
:i 12 —a
O
2, 10 ~~
:> 8 .-
a
a 4 _.
5° 2 «L-
o i i i i
10:7 10’6 10'5 10’4 10'3
CAPILLARY NUMBER

 

 

 

Figure 2.3. Relative effect of the capillary number as a function of trapped residual
saturation at Bond numbers less than 0.00667 (adapted from Morrow and
Songkran 1981).

l9

 

IRREDUCIBLE
——’ WETI‘ING

SATURATION
TRAPPED
RESIDUAL
N APL

k, SATURATION
_’ ‘—

nonwettin g
phase

 

 

 

 

0 —> S (wetting) 1

1 S (nonwetting) <— 0

 

 

 

Figure 2.4. Relative permeability as a function of phase saturation.

 

1 r

0.9 ..
0.8 a”
0.7 ..
0.6 ‘-
3 0.5 ‘r
ta? 0.4 ..
0.3 ‘-
0.2 ‘-
0.1 ‘-
0 i + . . .

0 0.2 0.4 0.6 0.8 1

kr

 

 

 

 

 

Figure 2.5. Flow passing through a permeable cylinder in a uniform flow field
relative to the flow passing through the clinder when k, = l (Wheatcraft and
Winterberg 1985).

20

 

 

0.9 r“
0.8 "
0.7 y-
0.8 ‘*

0.5 ‘1“

an/STI-n

0.4 «r

0.2 «r-

o.1 ..

 

 

0 i i i v

10‘6 105 Hr4 I0’3 10'2

CAPILLARY NUMBER

 

 

 

Figure 2.6. Relative correlation of the mobilization of residual oil and the capillary
number for uniform glass beads (adapted from Morrow et al. 1988).

 

  
  
   

HYDROPHOBIC
. 15:, , 1.3:: 523:2. ND

HYDROPHILIC
535:3?" END

IMMISCIBLE
CONTAMINANT

 

 

 

Figure 2.7. Schematic of a surfactant micelle capturing an idealized NAPL ganglia.

21

 

Critical
Concentration

    
     

Equlvalent
conductivity
Interfacial

 

 

 

 

CONCENTRATION (2.)

 

 

 

Figure 2.8. Relative changes in physical properties in the neighborhood of the
critical micelle concentration (adapted from Preston 1948).

22

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

SOIL s,” (%) REFERENCE
OTTAWA 12 POWER ET AL. 1992
SEVILLETA 27.1 WILSON ET AL. 1990
TRAVERSE CITY SOIL 15.8-19.3 WILSON ET AL. 1990
LLANO SOIL 13.9-18.5 WILSON ET AL. 1990
GLASS BEADS 14.25 CHATZIS ET AL. 1983
SANDSTONE 30-40 MELROSE AND BRANDER 1974
SEVILLETA 23-34 CONRAD ET AL. 1987
SANDSTONE 11-30 ' CHATZIS ET AL. 1988
SANDSTONE 27-43 CHATZIS AND MORROW 1984
OTTAWA SAND 15.9-20.4 PENNELL ET AL. 1993
WAGNER 18 SAND 8.4 POWERS ET AL. 1992
WAGNER 50 SAND 12.3 POWERS ET AL. 1992
OTTAWA SAND 12 POWERS ET AL. 1992
WAGNER MIX 16.5-18.5 POWERS ET AL. 1992
SEVILLETA SAND 22.5 MACE AND WILSON 1992
SAND/SILT/CLAY 16.9-30.3 MACE AND WILSON 1992
AQUIFER SAND 34-41 ANG AND ABDUL 1991
GLASS BEADS 121-14.06 MORROW AND SONGKRAN 1979

 

Table 2.1. Overview of past research in NAPL entrapment and mobilization.

 

Chapter 3
Materials and Methods

This chapter presents an overview of the materials and methods used in the
experiments presented in Chapters IV and V. A majority of the experimental work is
conducted in 1-D glass columns using white Ottawa sand, dodecane and a 5% solution of
a nonionic surfactant, Witconol NP-150, although some batch experiments used in the
early stages of investigation are also described. Methods are outlined for packing and
saturating the soil column, establishing a trapped residual NAPL, measuring permeability,
determining the equilibrium solubility of NAPL in surfactant, measuring the kinetics of
surfactant enhanced dissolution and dye sorption measurements. A section addressing

error analysis is also presented.

3 .1 Apparatus

A borosilicate glass column designed to minimize sorption losses and to maintain
a hydrophilic surface (Zalidis et al. 1991) was modified for use in this research (Fig. 3.1).
The column was altered for saturated soil experiments by removing the porous pressure
plate fiom the bottom of the column. In addition, the experiments conducted in saturated
soil do not require a venting port in the top cap. The column was designed for easy
assembly and disassembly. The length of the threaded rods can be changed to
accommodate different column lengths. In addition, Viton® o-rings, stainless steel
screens and Teflon® caps were used to maintain hydrophilic surfaces which minimize

sorption losses.

23

24

3.2 Preparation of Soil Column

Prior to packing the 1-D column, the individual components are washed and oven
dried. The total weight of all components was measured for later determination of soil
and fluid mass (Fisher Scientific Top Loading Balance Model # XT1200DR). The bulk

density, Pb, was determined from the dry weight of the assembled column.

"Mafia
V

6'

p1» (3-1)

where Md is the mass of the column with dry soil, MC is the mass of the column, and V0 is
the volume in the column.

The soil packed column was subjected to a 700 mm Hg vacuum (Vacu/Trol Lab
Vac. Regulator with pump, Spectrum Medical Industries, Inc., Patent Pending) for 30
minutes to remove the gaseous phase. The column was then slowly saturated fi'om the
bottom with deaired deionized water. The column was weighed to determine the

saturated weight of the column to estimate the pore volume (PV) and porosity (n) of the

soil.
PV= Mi'Md —V (3.2)
p caps
n = g (3.3)

where M, is the mass of the water saturated soil column and Vcaps the volume in the caps.

. 3.3 Establishing a Trapped Residual Irnnriscible Phase

The following procedure outlines the method for establishing a residual NAPL in
the soil column. Dodecane was flushed through the top of the water saturated column to
assure a stable displacement of the aqueous phase (Kueper and Frind 1988) under

approximately 1 psi of pressure. Once the NAPL reached the bottom of the column (as

25

evident through a liquid color change observed in the effluent), the flow of water out of
the bottom of the column was interrupted, while the pressure remained on the oil. After
1/2 hour, the aqueous flow was reintroduced through the bottom of the column using a
syringe pump (Harvard Apparatus Syringe Infirsion Pump 22, Model # 55-2219)
equipped with 140 cc Monoject® syringes. The effluent was monitored for free phase
NAPL. Trapped residual NAPL was observed in the soil pores near the column wall in
regions where the flushing water removed continuous NAPL. The column was flushed
with a minimum of 5 additional pore volumes of water after free NAPL was no longer
observed in the effluent to assure that all mobile continuous NAPL was removed,
simulating naturally occurring aquifer conditions which would trap NAPL as a residual in
the saturated region of an aquifer. When the column was disassembled, the residual
trapped NAPL was observed to be relatively uniformly distributed throughout the soil.
Flux rates of the infiltrating water were varied but maintained within the range of
velocities to assure laminar flow in the soil columns. In porous media, the upper limit of
Darcy flow to assure laminar flow is achieved by maintaining the Reynolds number (Re)
between 1-10, although Re = 100 is often mentioned as an upper limit (Bear 1979; Freeze
and Cherry 1979).

pvd
,u

 

Re = (3.4)

where p is the density of the fluid, v the velocity, d a characteristic length (e. g., diameter
of the soil grain), and u the viscosity of the fluid.

The saturation of trapped residual NAPL in the porous media, Sm, was estimated
to be (Wilson et al. 1990)

S... = M__M_ (3.5)
Ap PV

26

where Mm is the mass of the soil column after a trapped residual NAPL saturation is
established.

To simulate conditions in the upper region of the capillary fiinge, a trapped
residual gaseous phase was established by flooding a water saturated column with the
gaseous phase under 5 psi of pressure, followed by a minimum of 5 pore volumes of
water. An additional trapped residual NAPL was established in these columns to simulate
conditions in the upper region of the capillary fiinge by following the previously outlined

procedure for trapping residual NAPL in water saturated pores.

3.4. Hydraulic Conductivity Measurements
Hydraulic conductivity, K, was measured using the falling head method (Klute

1986) for initial experiments where

K =(%,)1n(1%2) (3.6)

where a is the cross sectional area of the standpipe, A the cross sectional area of the soil
sample, 1 is time, L is the sample length and H1 and H 2 measurements of the initial and
final hydraulic head (Fig. 3.2). For the experiment where the relative permeability of the
aqueous phase was measured during continuous surfactant pumping, conductivity was

measured by the pressure difference across the column using a water manometer.

19$

AAH (3.7)

where Q is the flux rate, L the length of the column and AH the hydraulic head difference
across the column (Fig. 3.3). Initial experiments were conducted to measure and account

for the head loss across the apparatus without soil.

27

3.5 Equilibrium Solubility of NAPL in Surfactant

Batch experiments where used to determine the equilibrium solubility of dodecane
over a range of surfactant concentrations in an aqueous solution. Ten [.11 of dodecane
were injected into 10 ml of surfactant in glass vials with butyl rubber/PTFE septa and
aluminum caps. The vials were placed on a shaker apparatus for 24 hours after which the
fluid contents was allowed to stratify for 24 hours in the vials with the septa side facing
down. Samples were taken from the bottom of the vials and analyzed for dodecane

concentration.

3.6 Investigation of the Kinetics of Surfactant Enhanced Dissolution of
Trapped Residual NAPL

To investigate nonequilibrium effects of surfactant enhanced dissolution, a
residual NAPL phase was initially established in a saturated soil column. A 5% surfactant
solution was pumped into the column at a flux of 10 ml/min. This faster perculation rate
maintained a minimal dodecane concentration at the initiation of the pulsed pumping
while assuring 1 < Re < 10. After flushing the column with 2 pore volumes of surfactant,
the pumping was discontinued. Surfactant was allowed to remain in static contact with
the residual NAPL for intervals ranging from 1-72 hours. At the end of the specified
interval, pumping was resumed, collecting 5 ml samples at the effluent end after 0.25 pore
volumes. After pumping 2 pore volumes of fresh surfactant into the column, the pumping
was again discontinued for a specified period of time. Experiments were also conducted
to compare the solubility of trapped residual dodecane into the surfactant enhanced
aqueous phase when the contact time was achieved from continuous pumping, as
opposed to static contact, through, a contaminated region. In these experiments, the
surfactant contact was controlled by varying the pumping rate between 0.24-1 ml/min,

resulting in residence times from 2-10 hours.

28

3.7 Dye Sorption Measurements

To observe the immiscible fluid movement and entrapment in the porous media,
the NAPL was dyed with an oil soluble dye, Oil Red 0 biological stain obtained from the
Aldrich Chemical Company (Milwaukee, Wisconsin). In these experiments, the mass of
soil that was placed glass vials with a fixed volume of dyed oil was increased. The vials
were well mixed on a mechanical mixer for 24 hours. The concentration of dye in the
fluid was then determined by ultraviolet (UV) analysis using a Hewlett Packard 8452A

diode array spectrophotometer.

3.8 Error Analysis

Uncertainty was estimated using a finite divided difference approximation of the

Taylor series where the overall uncertainty interval, 6R, is calculated as

5R: {glggdsfr

n V:
{Elm + 6x.) — R<x.- )]2}

i=1

(3.8)

Ill

The following statistical analysis was used to estimate the uncertainty in the
measurement of NAPL concentration. A calibration curve was generated to predict an
area response on the GC from known concentrations of dodecane using linear regression.

The 95% confidence interval for the true mean of y was estimated from (Anderson 1987)

 

1, (X—YY
n .( X12
Ely—Ln—

 

y = b+mxi ’a/2_,,_2Sr.x (3.9)

where y is the area response measured by the GO and x is the concentration of dodecane

(ppm)-

29

In the experimental work, the concentration of dodecane was estimated from the
measured area. The 95% confidence interval for concentration of dodecane (x) predicted
fi'om the GO results (y) is determined from

(Lb-71 ‘

_ I, _S.
Y b /,,n2rx 1+l+ m 2
m m n ZXz—(ZX)
\ n

 

(

 

(3.10)

 

 

J

For quality assurance, a new calibration curve is obtained at thebeginning of each
run on the CC. as well as measuring all samples in duplicate. In addition, the repeated
measurement of a known standard is made every 10 samples for instrument calibration

verification (ICV).

3.9 Material

White Ottawa sand obtained fiom Soiltest (Lake Bluff, Illinois) was used in these
experiments (Fig. 3.4) to a model homogeneous soil such as that commonly found in
sandy aquifers. The bulk density of the soil used was measured (Sec. 3.2) at 1.76 :t 0.05
g/cm3 (:1: 2.7%); the low standard deviation indicates that column packing
reproducibility method was successful. The porosity of the packed soil used was
measured at 0.39 i 0.02 g/cm3 (3: 4%); the low standard deviation indicates that
saturation method does not trap air. Deionized deaired water was used in the soil
column and to prepare surfactant solutions. A nonionic surfactant, Witconol NP-150
(ethoxylated nonylphenol) from Witco (New York, New York), was used as received to
enhance the aqueous phase solubility of the NAPL.

Dodecane from Phillips 66 Company (A subsidary of Phillips Petroleum, Borger,
Texas) is used as a surrogate NAPL. Dodecane is naturally found in the paraffin fraction
of petroleum and consequently is a by-product of the petroleum refining industry.

Dodecane can also be found in municipal waste and waste water due to the general use of

30

petroleum oils and tars (Verschueren 1983). Dodecane was selected based on the
following desirable physical properties 1) the specific gravity of dodecane is less than one
(LNAPL); 2) very low aqueous solubility; and 3) low volatility (Table 3.1). The last two
properties are necessary to limit the mass transfer between the phases during the
experiment, thus allowing the research to focus on the enhanced dissolution of the NAPL
into the surfactant enhanced water.

Wilson et al. (1990) suggest that trapped residual NAPL saturation in the
saturated region under low capillary and bond numbers conditions should be essentially
the same, regardless of the organic liquid. In experiments measuring the trapped residual
saturations of gasoline, p-xylene, PCE, decane, Soltrol and kerosene, it was determined
than the texture of the soil and its heterogeneity generally have far more control on
trapping that the composition of the organic liquid.

Adsorption isotherm tests established that the oil soluble dye used to trace the
dodecane, a red biological stain obtained from Aldrich, does not sorb to the soil particles

(Fig. 3.5) using a wavelength of 354 nm in U.V. spectrophotometry analysis.

31

 

 

S
S
s a
S S I
h n 8 w swam
nle nn a X]
clam Gen 1 ed
wtc .n. 8 NC
SSS V0 n
M. F
all;

 

    

.55.. . I . I . . I. (22.7..
aaraaaanaryaaaasrvauw
. .

D=5.44 cm

 

 

inflow/outflow

port

 

 

 

i § I
z ‘—

 

868-2";

 

 

 

Figure 3.1. l-D Column

32

 

 

 

  
    

soil column

 

 

 

 

 

 

atmospheric
pressure

 

 

 

Figure 3.2. Schematic of apparatus used to conduct falling head hydraulic
conductivity measurements.

 

 

 

 

 

 

 

 

 

 

 

 

 

 

Q 22¢ V ' 'Datumz=0

 

 

 

 

Figure 3.3. General schematic of apparatus used to measure hydraulic conductivity
based on the head difference across the column.

 

U.S. STANDARD SIEVE NOS.
so n
O

388

PERCENT FINER

88888

.-
O

 

O

100 10 l 0.1 0.01 0.001
GRAIN SIZE (mm)

 

 

 

Figure 3.4. Grain size distribution of white Ottawa sand.

 

 

 

 

OIL RED 0
0.4 ..
0.35 -+
0 9 . o v e
03 [ o . ' o ' o
E 0.25 l
fa”
E 02 l
< 0.15 »
3
0.1
0.05
0 . A . t 4 %‘
0 2 4 6 8 10 12
SAMPLE

 

 

 

Figure 3.5. Adsorption isotherm for Red-O dye.

34

 

 

 

 

 

 

 

Property Source

Solubility in pure water 3.7 ppb at 25° C Verschueren 1983
Surface tension 24.9 dynes/cm at 25° C Johnson and Dettre 1966
Interfacial tension with water 52.8 dynes/cm at 25° C Johnson and Dettre 1966
Density 0.749 g/cm3 25° C Measured

Interfacial tension with 5% Witconol 3.73 dynes/cm at 25° C Pedant Drop Method

NP-ISO

 

 

 

Table 3.1. Fluid properties of dodecane.

 

Chapter 4

Column Experiments in Surfactant Enhanced Dissolution of
Trapped Residual Dodecane

4.1 Introduction

Once released into the groundwater system, a nonaqueous phase liquid (NAPL)
will migrate through the vadose zone under the influence of gravitational forces.
Capillary forces will cause a continuous residual saturation of NAPL to remain in the
pore spaces. If the volume of NAPL is sufficiently large, it will reach the capillary fiinge.
NAPL more dense that water (DNAPL) will continue to move through the saturated
region until it is either completely exhausted as a discontinuous trapped residual phase in
the pore space or pooled at an impermeable boundary. NAPL less dense than water
(LNAPL) will pool at the capillary fiinge, held above the water saturated pores by
buoyancy and capillary forces.

A pool of LNAPL localized at the capillary fiinge can become smeared over a
larger region as isolated ganglia or blobs due to local pumping, NAPL migration or
seasonal water table fluctuations (Hunt et al. 1988; Conrad et al., 1992). The ganglia are
immobile and discontinuous pockets of NAPL trapped within the soil pores as a result of
capillary forces stronger than the combined viscous and buoyant forces. Trapped residual
NAPL can occupy between 10-50% of the pore space (\Vrlson and Conrad 1984).
Although the mass of NAPL in the pore space is reduced due to this migration, the
trapped residual NAPL serves as a continual source of groundwater contamination due to
the partitioning of NAPL components into the aqueous and gaseous phases. On average,

the maximum contaminant level (MCL) for typical NAPL contaminants are more than 2

35

36

orders of magnitude less than their solubilities (Miller et al. 1990). As a result, NAPL
dissolution into the groundwater can contaminate an aquifer for periods of time on the
order of decades, producing a waste stream of massive volume (Hunt et al. 1988).

Aqueous phase surfactants can be injected into the groundwater in an effort to
remove the NAPL by enhancing the solubilization of NAPL into the aqueous phase as
colloidal size clusters known as micelles. Once trapped in the micelle, the NAPL moves
in the same phase as the aqueous phase (Rosen 1979). Thus, surfactants make it possible
to increase the solubility of NAPL into the aqueous phase, decreasing the period of time
in which the trapped residual NAPL will contaminate an aquifer. Using conventional
pump-and-treat technology, the additional expense of surfactants requires cost-effective
and efficient controls for delivering the treatment to contaminated regions in the aquifer
in addition to recovering contaminant laden surfactant.

The use of the term equilibrium by hydrologists implies that the maximum
solubility of NAPL into the surfactant enhanced groundwater will occur almost
instantaneously. Nonequilibrium may result in less than predicted NAPL removal rates,
resulting in increased costs, if the time needed to achieve the solubility concentration of
NAPL into the surfactant is not achieved between the injection and recovery wells. In
contrast, continuing to pump surfactant which has achieved the solubility concentration of
NAPL through a polluted region is justifiably as inefficient.

Mass transfer of trapped residual NAPL into the surfactant enhanced aqueous
phase may also be limited by the variability in the shape of these discontinuous NAPL
blobs or ganglia. In the complex network of pore spaces within the soil, trapped residual
NAPL may be isolated as a single blob or branched through several pore spaces. As a
result, the solubility of NAPL into the surfactant enhanced groundwater may be limited at
the microscopic scale by dead end pOre spaces or groups of blobs blocking flow and

reducing the contact area for mass transfer.

37

4.2 Objective

The overall objective of this study was to evaluate surfactant effectiveness in
enhancing the aqueous phase solubility of NAPL. The specific objectives were 1) to
develop an analytical technique for measuring the concentration of NAPL in an aqueous
phase surfactant solution; 2) to compare the enhanced solubility of NAPL measured in
batch experiments with that observed when the NAPL is trapped as a residual in a porous
media; 3) to measure the breakthrough of surfactant injected into soil with trapped
residual NAPL; 4) to investigate non-equilibrium effects of surfactant enhanced solubility
of trapped residual NAPL; and 5) to monitor the dissolution of trapped residual NAPL
into a surfactant enhanced aqueous phase over time to evaluate the effects of variable
ganglia size and distribution on mass transfer for both continuous and intermittent

pumping.

4.3 Background

Immiscible fluids, such as NAPL, in porous media migrate under the influence of
capillary, viscous and buoyancy forces. However, in an aquifer at typical aquifer flow
rates, capillary forces are often the dominant force (Wilson et al. 1990), influencing the
variability in fluid saturations, permeability and trapping within the pore spaces. In a
NAPL spill in a sandy aquifer, typically hydrophilic, the aqueous phase is the wetting fluid
and NAPL the nonwetting fluid. In aqueous-gaseous phase systems, the bubbling point
pressure, or air entry pressure, has been defined as the pressure needed to force the
gaseous phase into a capillary tube while draining the water out. A similar definition
applies to NAPL displacing the aqueous phase. The NAPL entry pressure is the pressure
needed to displace the aqueous phase held in the soil pores by capillary pressure, PC (Eqn.
2.1). Since PC is inversely proportional to the radius of the pore, r, a higher NAPL entry
pressure is needed to displace the aqueous phase held in the smaller pores and pore

throats. As a result, NAPL penetrating a region where the aqueous phase saturates the

38

pores will displace the aqueous phase from the larger pores first. However, when the
aqueous phase is reintroduced into the pores, the stronger capillary forces in the smaller
pores and pore throats will pull the aqueous phase into these pore spaces first.

This phenomenon is illustrated in the mechanisms of snap-off and by—passing
developed in the singlet and doublet models used by petroleum engineers to explain the
trapping of residual oil in petroleum reservoirs. Wardlaw (1982) and Chatzis et al.
(1983) describes the concept of a singlet blob trapping, or snap-off, as a ratio of the pore
diameter to pore throat, which is referred to as the aspect ratio. The pore geometry with
a high aspect ratio will snap-off a discontinuous NAPL blob from the continuous flowing
NAPL being advected through the pore spaces by the infiltrating aqueous phase. In this
scenario, the small throat diameter reduces the capillary pressure between the two phases
abruptly and significantly enough to cause the snap-off of the NAPL. In contrast, in a
low aspect ratio the pore throat does not reduce sufficiently to alter the capillary pressure
as dramatically from one region to the next. As a result, the NAPL can move as a
continuous phase through the pore spaces.

Trapping can also be described through by-passing in a pore doublet model which
consists of a pore split into two pores, with one pore narrower that the other, and then
rejoined. When considering flow rates low enough to approximate aquifer conditions,
capillary forces are much larger than the dynamic viscous forces (Wilson et al. 1990). As
a result, capillarity will control the advance of the aqueous phase, which causes the

displacement of NAPL in the smaller pore space first (Chatzis and Dullien 1983). If the
. pore throat at the downstream connection is large enough to allow the NAPL-aqueous
phase interface to stabilize, NAPL will also drain from the larger pore branch. Otherwise,
the NAPL in the larger branch will break away from the continuous NAPL being
mobilized by the invading aqueous phase. A doublet model is often modified to include a
high aspect ratio within one of the branching pores to illustrate the possibilities of both

the by-passing and snap-off mechanisms of trapping NAPL (Chatzis et al. 1983).

39

The pore doublet model provides the basis for understanding the mechanism that
can trap a residual blob of NAPL over several pores. Trapped residual NAPL has been
cast in place in sandstone, glass beads and sand, illustrating the wide range of ganglia
sizes and shapes, by polymerizing styrene under high temperature and pressure (Chatzis
et al. 1983; Wilson et al. 1990; Powers et al. 1992). A single isolated drop in a pore
space can exist within the same region as elongated multiple pore ganglia or ganglia
branched throughout a complex pore network. Trapped ganglia can effectively plug a
constriction or cluster together to force fluid flow around a network of oil saturated
pores (Honarpour et al. 1986). In addition, the variability of ganglia shapes, sizes and
distribution may further limit the contact area available for mass transfer between the
NAPL and aqueous phase (Powers et al. 1992).

Past research has shown that the diffusion of NAPL into the aqueous phase is rate
limited (Weber et al. 1991; Powers et al. 1992; Borden and Kao 1989; Clarke et al. 1991;
Pennell et al. 1993). First order mass transfer rates can be estimated from a common
form of the mass transfer relationship for the dissolution of the NAPL into the aqueous
phase

dC
J=—=k C —C 4.1
dt 1( 3 ) ( )

This equation, which is Fick's first law, yields a linear driving force model for the
net flux of NAPL into the surfactant enhanced aqueous phase. J is a nonadvective flux
term which accounts for both molecular diffusion and hydrodynamic dispersion (Mercer
and Cohen 1990). The coefficient k, is the overall mass transfer coefficient which is a
function of a driving force and an interfacial area between the two phases of concern.
Miller et al. (1990) and Power et al. (1992) review theoretical and experimental work
which consider predicting 1:, based on varying surface area geometries, two—phase systems

and soil properties.

40

One such mass transfer model is based on the concept of a stagnant film model
(Levich 1962; Sherwood et al. 1975; Miller et al. 1990) where the general mass-flux

equation accounts for the thickness of the thin aqueous phase layer

J = k. (C. - C) = 95w. —— C) (4.2)

where D, is the free liquid diffusivity of the solute, and 6 is the thickness of the stagnant
film layer. This model assumes that the thin aqueous phase layer is stagnant. The term 6
is a function of D, (Miller et al. 1990). Miller et al. (1990) found that mass transfer rate
coefficients were directly related to aqueous phase velocity and NAPL saturation, but did
not establish a relationship to particle size. Calculations showed that their procedure
maintained the change in trapped residual NAPL saturation during an experimental run to
10% or less. Power et al. (1992) developed a phenomenological model to predict the
mass transfer using pore structure parameters as surrogate measures of specific surface
areas. As in Miller et al. (1990), Power et al. (1992) did not consider the impact of
interfacial area reduction, and subsequent decrease in mass transfer, due to dissolution
over time.

The use of intermittent pumping through the cycling Of injection/extraction well
pumping schedules is based on the concept of promoting the contact time needed when
the solubilization of the trapped residual NAPL into the aqueous phase is rate limited,
rather than instantaneous (Keely 1989). In addition, the control of the pumping rates and
location of the pumps with respect to natural gradients can provide additional hydraulic
controls which can be designed to optimize the contact time within the multiple
subsurface pathways within the capture zone of the extraction well (Palmer and Fish
1992)

41

4.4 Apparatus and Materials

A borosilicate glass column was designed for the surfactant enhanced dissolution
experiments to minimize sorption losses and to maintain a hydrophilic surface (Fig. 3.1).
The column was designed for easy assembly and disassembly. The length of the threaded
rods can be changed to accommodate different column lengths. In addition, Viton® 0-
rings, stainless steel screens and Teflon® caps were used to maintain hydrophilic surfaces
which minimize sorption losses.

White Ottawa sand obtained from Soiltest (Lake Bluff, Illinois) was used in these
experiments (Fig. 3.2) to simulate the homogeneous soil of a in sandy aquifers.
Deionized deaired water was used to wet the soil column and to prepare surfactant
solutions. A nonionic surfactant, Witconol NP-150 (ethoxylated nonylphenol) from
Witco (New York, New York), was used as received to enhance the aqueous phase
solubility of the NAPL. Technical grade dodecane obtained from Phillips Petroleum
(Borger, Texas) was used as a surrogate NAPL based on the following desirable physical
properties 1) the specific gravity of dodecane is less than one (LNAPL); 2) very low
solubility; and 3) low volatility. The last two properties are necessary to limit the mass
transfer between the phases during the experiment allowing the research to focus on the

enhanced dissolution of the NAPL into the surfactant enhanced water.

4 . 5 Methods

The methods for packing and saturating the column along with the steps used to
establish a trapped residual NAPL phase are outlined in Chapter 3. A literature review of
reasonable scope revealed only one analytical method for measuring dodecane
concentrations in a surfactant enhanced aqueous phase which incorporates C 14-1abeled
dodecane (Pennell et al. 1992). An analytical method to measure dodecane
concentrations. in a surfactant enhanced aqueous phase is needed which will allow

measurements in an open laboratory space as well as in the field. An analytical method

42

for measuring the concentration of dodecane in a surfactant enhanced aqueous phase has
been developed which utilizes gas chromatography (G.C.) with headspace sampling using
a Perkin Elmer PE Autosystem (600 Series, Model No. 1000) equipped with a flame
ionizer detector and a PEHS40 automatic headspace sampler. Headspace G.C. relies on
the partitioning of the analyte from the aqueous phase into the gaseous phase. A sample
is withdrawn from the gaseous phase to determine the analyte concentration by G.C.
analysis. Tests Show that the degree of partitioning of dodecane into the gaseous phase is
strongly influenced by surfactant concentration in the aqueous phase (Fig. 4.1). Since a
surfactant enhanced aqueous phase may become diluted by ambient groundwater,
assuming a 5% solution of surfactant may result in an inaccurate estimate of dodecane
removal. To prevent this, the analysis of aqueous phase samples would require measuring
both the surfactant and NAPL to determine concentrations based on two different sets
of standard calibration curves. To avoid this complication, experiments were conducted
to consider the use of salts, a surrogate NAPL, and dilution were evaluated.

The 5% surfactant stock solutions were prepared gravimetrically using deionized
water. The mixture was stirred for 24 hr. using a magnetic stirrer plate would . Lower
concentrations of surfactant solutions were prepared volumetrically by diluting the 5%
surfactant solution with deionized water. Due to the low solubility of dodecane in water,
standards of known concentration of dodecane were prepared by dissolving dodecane in
methanol and making subsequent dilutions in surfactant. ‘

The kinetics of the dissolution of dodecane into the surfactant enhanced aqueous
phase was studied using a flow interruption technique common to investigations of rate-
limited processes including intraparticle diffusion (Hellferich 1962), film diffusion (Weber
and Morris 1963) and sorption of hydrophobic Organic solutes by soils (Brusseau et al.
1989). The technique involves injecting surfactant into a soil column with an established
discontinuous trapped residual NAPL phase. The flow is then interrupted for a specified

period time to establish a contact time between the surfactant enhanced aqueous phase

43

and the trapped residual NAPL. At the end of the specified interval, the flow is
continued. Effluent concentrations are measured for the first 0.25 pore volumes
displaced. Flow is continued for 2 pore volumes in order to assure a fresh injection of
surfactant prior to interrupting the flow. Increases in the effluent concentration of NAPL

over time are indicative of rate-limited solubilization.
4.6 Results and Discussion

4.6.1 Analytical Method

The use of surfactants in the experimental work caused heavy sudsing and the
formation of an emulsion during both the liquid-liquid extraction and solid-liquid
extraction procedures. As a result, an analytical method was considered using head space
gas chromatography using a Perkin Elmer PE Autosystem gas chromatagraph equipped
with a flame ionizer detector and a PEHS40 automatic headspace sampler. In this
analytical method, the quantification of dodecane concentration in aqueous phase samples
is dependent upon the partitioning of the dodecane from the aqueous phase into the
gaseous phase. Tests show that the degree of partitioning is strongly influenced by the
surfactant concentration in the aqueous phase (Fig. 4.1). In these experiments, a range
of surfactant concentrations were prepared with a range of known dodecane
concentrations. The G.C. analysis indicates that a lower concentration of surfactant
results in a greater G.C. response than does a higher concentration of surfactant given the
same amount of dodecane. As a result, the slope of the calibration curve increases as the
concentration of the surfactant decreases. Using this methodology, one would need to
measure both the surfactant and dodecane concentrations in the sample. However, the
surfactant was not detectable using GC/FID. To avoid using a second analytical method
to measure the surfactant concentrations, the use of salts, a surrogate NAPL, and dilution

were evaluated.

44

Samples were saturated with NaCl in an attempt to salt out the dodecane and
lessen the differences due to variations in surfactant concentrations. Samples of various
surfactant concentrations were spiked with 4 ppm dodecane. Duplicate samples were
prepared by adding 1.5 g NaCl. The addition of salt resulted in a decrease in the
partitioning of the dodecane into the headspace, which did not overcome the problem
(Fig. 4.2). An additional set of samples were prepared over a range of surfactant
concentrations and spiked with 4 ppm dodecane and 4 ppm undecane. A linear relation
between surfactant concentration and the G.C. measurement of undecane would make
undecane an ideal surrogate for determining surfactant concentrations. However, the
measurement of undecane was also found to vary with surfactant concentration (Fig. 4.3).

It was determined that diluting samples 100:1 (100 pl sample in 10 ml surfactant)
in 5% surfactant would maintain a reasonably consistent surfactant concentration (4.95-
5%) for all samples, regardless of their initial surfactant concentration. Dilution also
provides a solution to free phase NAPL that may be released from the soil pores and
collected as a separate phase in the effluent as a result of lowering the interfacial tension
(IF T). If this free product NAPL is released at the same time that a sample is collected at
the effluent end of the column, a diluted sample can be immediately prepared from the
distinctly separate aqueous phase of the effluent for a measurement of solubilized NAPL.

The G.C. operating parameters (i.e., thermostat temperature, equilibrium time and
oven temperatures) were optimized for the measurement of dodecane (Fig. 4.4). The
higher thermostat temperatures resulted in a greater partitioning of dodecane into the
headspace, whereas the longer equilibrium time only slightly improved the partitioning.
The oven temperature was varied between 100 - 150°C to improve separation of peaks.
The optimum oven temperature was determined to be 125°C. Table 4.1 shows the final

setting selected for the G.C.

45

4.6.2 Equilibrium Solubility

Batch experiments were initially conducted to measure the equilibrium
concentration of NAPL over a range of surfactant concentrations for the dodecane-
Witconol NP-150 system studied. Three different batches of Witconol NP-150 were
obtained from Witco Corporation, which will be referred to as batches A-C, respectively
(G-6326, R-686, G-7971). Batch B proved more difficult than the other surfactant
samples to mix in deionized water, resulting in a milky color solution with an emulsion.
Because of this phenomenon, this batch was excluded from this research.

A 5% solution of Witconol NP-150 enhanced the solubility of dodecane into the
aqueous phase from 0.0035 ppm to 2005 ppm i 2%. Varying the surfactant
concentration from 0.01-5% resulted in enhanced solubilities that ranged from 12.05 to
2005 ppm (Fig. 4.5). The critical micelle concentration (CMC) is defined as the
concentration where micelles first begin to form. It can be estimated be breaks or
changes in various physical properties of the surfactant which occur when the micelles
begin to form. Figure 4.5 indicates that the CMC occurs at approximately 0.1%, which is
in agreement with the predicted CMC from the graph of interfacial tension versus
surfactant concentration (Fig. 4.6). Since the solubility of dodecane in water at 3.5 ppb
(0.0035 ppm), concentrations as low as 0.01% surfactant increase the solubility of
dodecane by several orders of magnitude. Based on a least squares fit to the slope of the
batch solubilization data above 0.5%, the following expression can be used to determine
the equilibrium solubility of dodecane, Cs (ppm), as a function of the concentration of
surfactant, Cm4(%)

C, =403.72C,,,, — 1.89 (4.3)

with an R2 value of 0.999.
Remediation design is often based on the assumption that equilibrium solubility is

achieved once the treatment is introduced in-situ. However, chemical and physical

46

processes inherent in the subsurface environment may inhibit equilibrium concentrations
or exhibit rate-limiting kinetics resulting in less than expected rates of clean-up. An initial
experiment was conducted to detemrine is the equilibrium solubility could be achieved in
the porous matrix. A 5% solution of Witconol NP-150 was injected into a soil column
with a uniform distribution of trapped residual NAPL. Once the column was flooded
with the surfactant, pumping was discontinued for 24 hrs. to allow sufficient contact time
between the surfactant enhanced aqueous phase and the residual NAPL phase in order to
determine whether or not equilibrium solubility (C s) could be achieved during this period
of time. In this experiment, the solubility was decreased to approximately 0.85C5,
indicating mass transfer limitations (Fig. 4.7). The error in the measurement of the
experimental data is shown in Figure 4.8. In a separate experiment, a 24 hr. no flow
period was also studied after a column was flushed with 90 pore volumes (pore volumes)
of surfactant. In this case, a much higher rate of removal, approximately 0.99C5, was
achieved (Fig. 4.9). This suggests that the size and shape of the ganglia and pore
restriction to flow due to the trapped immiscible phase present may change over time as
the surfactant enhanced dissolution of the ganglia reduces the volume of trapped NAPL.

These phenomenon will be investigated firrther in the following sections.

4.6.3 Kinetics of Dodecane Dissolution

The kinetics of the dissolution of dodecane into the surfactant enhanced aqueous
phase was studied using a flow interruption technique common to investigations of rate-
limited processes including intraparticle diffusion, film diffusion and sorption of
hydrophobic organic solutes by soils. Increases in the effluent concentration of dodecane
over time are indicative of rate-limited solubilization, as shown in Figure 4.7. The data
shown was collected from two separate column experiments, with an initial trapped

residual NAPL saturation of 17.6% and 16.1%. A 5% surfactant enhanced aqueous

47

phase was pumped through the column at a rate of 0.5 ml/min. The interrupted flow time
was varied between 1-24 hours.

It is important to consider that the mass transfer is a combined result of advective
and diffusive process as well as other potential processes such as chemical kinetic and
density driven flow (Miller et al. 1990). The first order mass transfer coefiicient was
estimated by lumping all of these processes as one coefficient in the following analysis.

The transport equation for the aqueous phase in 1-D accounts for dispersion (D),

advection (v), and dissolution (J) of the trapped residual NAPL in the aqueous phase.

 

2
£=Dd§lv£u (4.4)
dt dx dx
In the no-flow situation, transport is dominated by the dissolution of the trapped
residual NAPL, which can be modeled using first order kinetics incorporating a linear

driving force or lumped parameter model to estimate the rate limited process.

61C

(3).... = J = k,(C, — C) (4.5)

The mass transfer coefficient, k,, is a lumped parameter which accounts for the
mass transfer rate between the aqueous phase and the NAPL and the changing surface
area available for mass transfer. The ordinary differential equation can be solved the

separation of variables and integrating
C r
f L = j k,dt (4.6)
C, 0

Assuming that the initial concentration of NAPL in surfactant is zero (i.e. C0 = 0)

_ :(1_e-ktr) * (4.7)

1

48

The data from the flow interruption study can be rearranged to solve for k, using linear
regression.

C
k, r _ ln[1— C] (4.8)

S

This equation assumes that the data asymptotically approaches the equilibrium
concentration of NAPL. However, the data indicates that the maximum solubility in the
porous matrix is approximately 0.8Cs. Using the software Scientist (MicroMath
Scientific Software, Utah) to solve for the mass transfer coefficient, k, is 0.161 hr'l,
assuming C s is achieved. Accounting for a reduced maximum solubility of 0.8Cs in the
porous matrix, k, is 0.387 hrl (Fig. 4.7). This second value will be used in the design

stage of the research. Output from the software analysis is included in the appendix.

4.6.4 Breakthrough Curve

A 5% solution of surfactant was injected into a soil column contaminated with
immobile residual dodecane. The concentration of dodecane in the effluent was measured
as a firnction of time in order to develop a breakthrough curve. The initial pore volume
was detemrined from the weight of a soil column saturated with deionized water (Sec.
3.3.1). However, the trapped residual phase reduces the volume of water that would

flow in the column, requiring a refined definition of pore volume
PV' = (1 — S,,,)PV, (4.9)

where PV“ is the redefined pore volume based on the previous definition of the pore
volume, PV,, and the trapped residual NAPL saturation, Sm. This definition assumes that
the residual NAPL is evenly distributed throughout the porous media. It does not
account for dead end pore spaces due to trapped ganglia either effectively plugging a
constriction in the pore space or clustering together to force fluid flow around a network

ofNAPL saturated pores (Honarpour et al. 1986). The variability of ganglia shapes, sizes

49

and distribution may also limit the contact area available for mass transfer between the
NAPL and surfactant enhanced aqueous phases.

The breakthrough curve of surfactant injected into a region of trapped residual
NAPL was measured in four different column experiments. Initial conditions are
summarized in Table 4.2. Trapped residual saturations range between 131-18.0%. A
5% solution of Witconol NP-150 was pumped through the columns at a flux of 0.5
ml/min. Figure 4.10 shows the breakthrough of the surfactant injected into a region of
trapped residual NAPL for each column. In these figures, scaled time, r*, is determined

from

t =— (4.10)

where v is the average linear velocity, t is time and L is the length of the column.

Figure 4.11 summarizes the error associated with values of C/Cs. A compilation
of the 4 experimental runs is presented in Figure 4.12. The differences in the
breakthrough curves may be explained in part by variation in local velocities due to
variations in both trapped residual NAPL saturation and porosity as well as variable
ganglia sizes and shapes. As a result, the local residence or contact time between the
surfactant enhanced aqueous phase and the NAPL blobs may vary.

The combined effects of surfactant sorption and variable ganglia shapes and sizes
on the breakthrough curve are difficult to uncouple in these experiments. Edwards and
Luthy (1990) determined that in nonionic surfactants, micelles do not tend to sorb, but
that significant surfactant sorption does occurs below the CMC. Since a 5% solution of
surfactant was used, which is significantly above the CMC of 0. 1%, surfactant sorption to
the soil is assumed to be negligible.

‘ The final C/Cs is approximately 64% for all four experiments. The advected
contact time for these experiments is approximately 4.7 hrs. Using the mass transfer

coeflicient determined in the flow interruption study, C/CS is estimated at approximately

50

0.67. The breakthrough curves in general fall slightly below this value, which can be
explained by experimental error, variability in trapped residual NAPL saturation,
variability in ganglia shape and size and limitations in predicting the mass transfer using a

lumped mass transfer coefficient.

4.6.5 Long Term Continuous Pumping

The pumping was continued for approximately 90 pore volumes in order to
monitor the dissolution rate as a firnction of time (Fig. 4.9). The purpose of the long run
is to evaluate the mass transfer limits that may occur as the interfacial area is altered as a
result of changing ganglia Shapes and sizes due to dissolution. The results indicate that
the dissolution changes over the course of pumping, but with no distinct upward or
downward trend. The results may be interpreted as a result of the changing surface area
of the ganglia, which may be dynamic within a series of pore spaces. As a smaller ganglia
is reduced enough to move through a constricting pore throat, it will migrate and
coalesce with a. ganglia in an adjacent pore space (Ng et al. 1978). As a result, the
available surface area of the newly formed ganglia is altered, effecting mass transfer
locally. After 90 pore volumes, the column was subject to a clear water flushed. The
column was weighed in order to measure the remaining residual. Mass balance indicate
that the column was cleaned of trapped residual NAPL. A second injection of surfactant
was flushed through the column and allowed to remain static for 24 hrs. Results
indicated that NAPL was still in the porous matrix. In addition, NAPL was removed at a
higher normalized effluent than in the previous pulsed pumping. This can again be
attributed to the change in ganglia shape and sizes. The reduced ganglia may result in
larger surface area and smaller throat constrictions. The pulsed pumping was continue
for three more 24 hr intervals. The last pulse pumping indicated a reduction in NAPL

removal.

51

4.7 Summary

An analytical method for measuring the concentration of dodecane in a surfactant
enhanced aqueous has been developed for head space G.C. This method is dependent
upon the partitioning of the NAPL from the aqueous phase into the gaseous phase. Tests
show that the degree of partitioning is strongly influenced by surfactant concentration in
‘ the aqueous phase. To overcome the effect of surfactant concentration, samples are
diluted 100:1 in 5% surfactant. The detection limit for the G.C. measurement of
dodecane concentration in a surfactant enhanced aqueous phase is approximately 10 ppb.

Column experiments were conducted to evaluate the use of surfactants to enhance
the dissolution of trapped residual NAPL below the capillary fiinge. Results from
intermittent and continuous pumping of a nonionic surfactant, Witconol NP-150, into a
region of trapped residual dodecane encourage further research in the use surfactant
enhanced aquifer remediation may be a viable in-situ remediation technology.
Equilibrium batch solubilization experiments estimate that a 5% solution of Witconol NP-
150 in water can enhance the solubilization of dodecane by 5 orders of magnitude. In the
porous matrix, 80% of the equilibrium solubility can be achieved after 12 hrs of contact
time, regardless whether the contact time was established through pulsed or continuous
pumping. Breakthrough curves were measured for 5% surfactant injected into a region
of trapped residual NAPL which did not strongly indicate sorption of surfactant to the
soil. A continuous pumping experiment indicated that the area available for mass transfer
varies over time. A final clear water flush followed by pulsed pumping of surfactant
significantly improved NAPL recovery after an initial 90 pore volumes of surfactant had
been pumped through a region.

Further investigation will be conducted to evaluate the use of hydraulic controls
to promote the delivery of surfactant into regions near the capillary fiinge contaminated

with trapped residual NAPL.by incorporting modeling in the 2-D in-situ environment.

52

 

 

 

 

 

 

 

1.8 --
16 0 ........ <> ........ 5% ,4
1.4 __ A 2.50%
A ‘— ‘ o i
3 1.2 1.25%: D
E 1 - D 0.60% ,.4' D
:5 0.8 —~ --------- ., ------ 0.30% I
E 4 U A
< 0.6 -- " ‘
04 D A A
__,4 A
02 — -- a ‘ A ...................... 0
A IIIIIIIIIII é ‘‘‘‘‘‘‘‘‘‘‘‘‘‘‘‘‘‘‘‘‘‘‘‘‘‘‘‘‘‘ O ............................................
0 ............... 9f . . { ' e .
0 1 2 3 4 5 6 7 8 9 10
Cone. of Dodecane (ppm)

 

 

 

Figure 4.1. GC response to dodecane concentration at varying surfactant
concentrations illustrating the effects of surfactant concentration on partitioning of
dodecane.

 

 

—D— DODFCANE W/O SALT

—‘°_ WDFCANE W/SALT

 

 

 

 

AREA/le6(uV‘see)
c88888888

 

A . . . _ _
05 r 1.5 2 25 3 3.5 4 45 5
Corie. d'WIfcaId Nit-150w.)

O

 

 

 

Figure 4.2. GC measurement of 4 ppm dodecane in varying surfactant
concentrations with salt.

53

 

 

 

 

 

 

 

Cone. of Witconol DIP-150 (%)

 

la) --
160 *-
—C}—-

3‘ 140 “ DODBCANE
5 1m .. "" A WANE
5 1(1) ..
3 30 ..
2
g m 4%
< 40 ..

20 ..

0 . . . . r . . t fl:
0 0.5 l 1.5 2 2.5 3 3.5 4 4.5

Ur

 

Figure 4.3. Four ppm undecane at varying surfactant concentrations.

 

 

 
 
     
 

 

 

 
 

 

   

 

 

 

 

12

160 --
I” «h —0— 95/30 /
_.x__
m -- ”/60 R=o.992 Y=14.14X+8.6
’3‘ —0— 80/30
8 ‘00 ‘" R=0.998 Y=15.00X+2.2
V
‘8 80 ‘-
- R=0.998 Y=6..17X+19
a 60 . // //
‘0 T // R=0.995 Y=6.0.8X+17
m ..
0 JF// , 4 a
o 2 4 6 s 10
Cone. of Dodecane (ppm)

 

Figure 4.4. The influence of thermostat temperature and equilibrium time
(temp/time) on calibration curves.

 

 

54

 

 

 

 

 

2500 l 2500 ..
2000 l 0 2000 0
3 A (‘3 A
E g 1500 4 E 5 1500 4
J v _’ V
g ‘i‘ g ‘5
.1 5 1000 . 0 O 5 1000 -. 0
00) “J a) u’
3 § 500 " 0 3 § 5‘” «1» O
0 O
0 -———<)—0—<>—A ne- 9 O 9 a O @O t c . . .
0.001 0.01 0.1 1 10 o 1 2 3 4 5
as SURFACTANT (WITCONOL NIP-150) ‘56 SURFACTANT (WITCONOL NP-150)

 

 

 

 

Figure 4.5. Equilibrium solubility of dodecane in varying concentrations of Witconol
NP-150.

 

8

8 8

 

 

N
C

Surface Tension (dynes/cm)
8 ‘5

 

WWW—LAM

O -
0.0001 0.001 0.01 0.1 1 10
Cone. of Witconol PIP-150 (%)

 

 

 

 

Figure 4.6. Critical micellar concentration of Witconol NP-150 determined from
surface tension (Witco 1991).

 

55

 

 

 

 

 

 

 

 

 

1.0-
08-
06-
on
a 04-
6 Raw Data
-——m——— :: '1
02, k I 0.387 hr
—1t,=0.161hr'l
00-
, 1 1 r I T ' T ' I
o 5 10 15 20 25
Residence Time (hr.)

 

 

 

Figure 4.7. Flow interruption study.

 

16- 4

141--

12'-

Relmive Error (%)
6

 

4 A A A
00 02 O4 08 .08 1.0

 

 

 

Figure 4.8. Error associated with enhanced solubility measurement in flow
interruption study.

56

 

 

 

1 l . O O

0.8 «» °

4’ 4

.1 0.6 ’4. 4 ’ 4., .° 0 ° , °
2 . .. O . O
U 0.4 ° 0 N

0.2 . sTOP CONTINUOUS PUMPING

0 ¢ 4 4 t t
0 20 40 60 80 100 120
t*

 

 

 

Figure 4.9. Continuous pumping of Witconol NP-150 followed by a water flush and
subsequent pulsed pumping.

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

1.00 .. 1.00 l
0.80 .. 0.80 t»
g 4 4 4 ° 0
44 , . 4 4 4 9 O . 4 °
a 060 .. 4" .. 0.60 .. ”
O 4 U
D 040< ° O 040 °.
' . Srn=14.8'/t ' Sm =13.1°/.
0.20 ’ 0.20 t .
4 .4
0m m‘“ “““‘ "52“.: 4 4 4 0m --—4+.-o—++M—4———4———4
0000 0.500 1.000 1.500 2000 2500 0.000 0.500 1.000 1.500 2.000
t. t O
100 1.00
0.80 1» 030 ..
g 400. 0 . 4
n 0.60 .. ”4.. ° ° 0 ° a 0.60 -- 4’ °
0 4 Q 4
U 4
0.40 <- 0 040 ..
020- 5m: 160% 020 .0 Sm -18.0%
0 0
0m a———O-O-O+M——4——+———4——t 0m «t—W—OW—WM‘Q—fi—t—fl
0.000 0.500 1.000 1.500 2000 2500 0.000 0.200 0.400 0.600 0.800 1.000 1.200 1.400
t. t.

 

 

 

 

 

 

Figure 4.10. Breakthrough curves for 5% Witconol NP-150 injected into regions of
trapped residual dodecane.

57

 

 

 

 

 

 

 

 

12 7'
0
10 it
0 13.10%
M
O 8 -t>—
g ‘ 14.80%
E 6 A 16.00%
a 4 0 18.00%
0
O
2 a!
a
0A
O
0 iii 9 4 at tam—WW
0.00 0.10 0.20 0.30 0.40 0.50 0.60 0.70
C/Cs

 

 

 

Figure 4.11. Error associated with break through curves.

 

 

 

 

 

 

 

LN 1'
0.” 1r
°Sln=11196 ‘ o 88
0° RfiAfiédAa‘ ‘ A ‘ ‘ ‘ A
0.50 .. A Srn=14.8% geinA a A
A; ‘
8 ASrn=16.0% A
.0
0.404 ostn=raosi A ° ‘
‘
O
0.1)» 4
000A
0
am-i—MWQA‘; 1 $ %
0.” 0.51!) HID 1.5m . 201) 2.5“)

t.

 

 

 

Figure 4.12. Compilation of breakthrough curves.

58

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

Sample temperature 95°
Needle temperature 120°
Transfer temperature 120°
G.C. cycle time 10 minutes
Thermostat time 30 minutes
Pressurized time 2 minutes
Inject time 0.15 minutes
Withdraw time 0.2 minutes
Oven temperature 125 ° C
Carrier gas (Helium) 200 ml/min
Table.4.1. G.C. Settings.
Exp. Sm Relative Relative pb Relative
Error (%) Error (%) Error (%)
1 0.148 2.533 0.376 0.484 1.766 0.058
2 0.160 2.427 0.401 0.479 1.803 0.058
3 0.131 2.668 0.369 0.486 1.854 0.058
4 0.180 2.381 0.383 0.483 1.840 0.058

 

 

 

 

 

 

 

Table 4.2. Experimental set-up for breakthrough curves.

 

Chapter 5

Aqueous Permeability Prior to and During Surfactant Enhanced
Dissolution of Trapped Residual NAPL

5 .1 Introduction

Once introduced into the groundwater environment, a spill of nonaqueous phase
liquid (NAPL) will migrate through the vadose zone of the soil. If the volume is
sufficiently large, the spill will reach the capillary fiinge where isolated ganglia of NAPL
can become trapped as a discontinuous residual phase. Once discontinuous, the NAPL
becomes a continual source of groundwater contamination that is not easily addressed by
conventional pump-and-treat remediation technology (Hall, 1989). Removal is only
possible by the slow dissolution of the NAPL into the aqueous phase, since the NAPL is
essentially immobilized by capillary forces stronger than the combined viscous and
buoyancy forces. On average, the maximum permissible level of a NAPL contaminant
acceptable in water is defined by the US. Environmental Protection Agency as more than
2 orders of magnitude less than the solubility of typical NAPL contaminants (Miller et al.
1990). As a result, NAPL dissolution into the groundwater can contaminate an aquifer
for decades, producing a dilute waste stream of massive volume (Hunt et al. 1988).
Aqueous surfactants can be injected into the groundwater to enhance the solubilization of
NAPL into colloidal size clusters known as micelles. Once trapped in the micelle, the
NAPL moves in the same phase as the aqueous phase (Rosen 1989). The added expense
of surfactants to pump-and-treat methods accentuates the need for efficient delivery of

the treatment into contaminated regions.

59

The limited number of field applications to date of in-situ remediation
technologies for the treatment of non-aqueous phase liquids (N APL) have not achieved
the high efficiency suggested by 1-D column experiments (Ellis et al. 1984; Nash and
Traver 1984; Nash 1985). This is due in part to the fact that 1-D columns impose an
artificial boundary which controls fluid movement between injection and recovery points.
However, in the field flow is free to by-pass the low conductivity NAPL contaminated
region due to the lack of confinement of flow boundaries that coincide with the
boundaries of the contaminated region. Since many in-situ technologies rely on transport
through the aqueous phase, the reduced aqueous conductivity in. the contaminated region
may result in some fraction of the treatment by-passing the area targeted for remediation.
This resistance can be decreased if the NAPL saturation is reduced through pumping,
NAPL migration or, as in the case of LNAPL, spread over a larger range due to
watertable fluctuations (Hunt et al. 1988). The remaining residual saturation is far more
difficult to remove, trapped by capillary forces stronger than the combined opposing
viscous and buoyant forces. In the saturated region, residual NAPL occupies between
10-50% of the pore space (Wilson and Conrad 1984). Hence, the pore blockage resisting
flow may still be significant. The problems caused by blockage may be accentuated in the
capillary fiinge, where a trapped residual gaseous phase is also trapped within the soil
pores due to imbibition and drainage in the capillary fiinge from infiltration and water
table fluctuations (Corey 1984).

Relative permeability-saturation curves are used to predict the flow of two
immiscible fluids within a porous media. Within the matrix, the wetting phase fluid is
physiwa limited fi'om reaching firll saturation or its maximum relative permeability
within the pores if the non-wetting phase is trapped and discontinuous (Fig. 2.4).
Petroleum engineers want to mobilize trapped residual oil, which is the nonwetting fluid
with respect to the aqueous phase, which remains after a normal waterflood in petroleum

reservoirs. As a result, past petroleum engineering research has focused measuring the

61

relative permeability of the aqueous phase at Sa 2 1 - Sm. in sandstone cores (Morrow et
al. 1985; Gilliland and Conley 1975; Amaefiile and Handy 1981) and glass beads
(Morrow and Songkran 1981). In these experiments, the saturation of the NAPL was
reduced by lowering the interfacial tension between the aqueous phase and the NAPL or
by increasing the pressure gradient. It has been shown. that the pressure gradients
required to mobilize trapped residual NAPL in aquifers is unrealistic (Wilson and Conrad
1984). Mobilizing trapped residual oil may cause the oil to migrate into previously
uncontaminated regions of the aquifer, a problem accentuated in the remediation of
DNAPL. However, experiments measuring the aqueous phase permeability in regions
over a range of trapped residual NAPL saturations may provide design values germane to
developing remediation strategies. However, the effect of the interfacial tension on the
relative permeability curves of the aqueous phase has been evaluated with conflicting
results. Gilliland and Conley (1975) extend the curve for the relative permeability of the
aqueous phase from Sa 2 1 - Sm to S8 = 1 without providing details on how the
composite data was acquired (Morrow et al. 1985). The work of Amaefule and Handy
(1981) in contrast shows that the aqueous phase relative permeability curve changes as
the interfacial tension decreases. Similar work conducted by Morrow et al. (1985)
showed that reduction in the interfacial tension by an order of magnitude produced no
obvious change in the relative permeability. To the knowledge of the author, similar
measurements have not been made in unconsolidated natural soils for aqueous phase
permeability as a function of trapped residual NAPL or gaseous phase saturations,
especially when the residual NAPL saturation is reduced as a result of mass transfer into
the aqueous phase.

The overall objective of this research was to develop a better understanding of the
relative permeability of the aqueous phase characteristic of a region near the capillary
fringe contaminated with residual trapped NAPL before and during surfactant treatment.

The specific objectives were 1) to measure the saturation of trapped residual dodecane in

62

a naturally occurring unconsolidated porous media as a function of the ratio of viscous
and capillary forces; 2) to measure the relative permeability of the aqueous phase in the
media as a function of the saturation of trapped residual NAPL; 3) to measure the
reduction in relative permeability of the aqueous phase as a function of two trapped
residual immiscible phases (NAPL and gaseous phase) simulating conditions that may
occur in the capillary fiinge; and 4) to measure the increase in the relative permeability of
the aqueous phase as the trapped residual NAPL is reduced by surfactant enhanced

dissolution.

5.2 Materials and Methods

A borosilicate glass column designed to minimize sorption losses and to maintain
a hydrophilic surface (Zalidis et al. 1991) was modified for use in this research (Fig. 3.1).
The column was altered for saturated soil experiments by removing the porous pressure
plate from the bottom of the column. In addition, the experiments conducted in saturated
soil do not require a venting port in the top cap. The column was designed for easy
assembly and disassembly. The length of the threaded rods can be changed to
accommodate different column lengths. In addition, Viton® o-rings, stainless steel
screens and Teflon® caps were used to maintain hydrophilic surfaces which minimize
sorption losses. White Ottawa sand obtained from Soiltest (Lake Bluff, Illinois) is used
in these experiments (Fig. 3.2) to model homogeneous soil commonly found in sandy
aquifers. Deionized deaired water is used to wet the soil column and to prepare
surfactant solutions. A nonionic surfactant, Witconol NP-150 (ethoxylated nonylphenol)
used as received fi'om “frtco (New York, New York), was used to enhance the aqueous
phase solubility of the NAPL. Dodecane from the Phillips 66 Company (Borger, Texas)
is used as a surrogate NAPL based on the following desirable physical properties: 1) the
specific gravity of dodecane is less than one (LNAPL); 2) low toxicity; 3) very low

solubility; and 4) low volatility. The last two properties are necessary to limit the mass

63

transfer between the phases during the experiment consequently allowing the research to
focus on the enhanced dissolution of the NAPL into the surfactant enhanced water.

To simulate conditions in the upper region of the capillary fringe, a trapped
residual immiscible phase was established by flooding a water saturated column with
approximately one pore volume of the immiscible phase under 5 psi of pressure, followed
by a minimum of 5 pore volumes of water. In some of the columns, a trapped residual
gaseous phase was established followed by a trapped residual NAPL phase to simulate
condition near the capillary fiinge. A thorough discussion of these methods is presented
in Chapter 3.

Hydraulic conductivity, K, was measured using the falling head method (Klute

1986) for initial experiments

K = (%,)1n(%2) (5.1)

where a is the cross sectional area of the standpipe, A the cross sectional area of the soil
sample, 1 is time, L is the sample length and H, and H 2 measurements of the initial and
final hydraulic head (Fig. 3.2). For the experiment where the relative permeability of the
aqueous phase was measured during continuous surfactant pumping, conductivity was

measured by the pressure difference across the column using a water manometer

K = -—QL

AAH (5'2)

where Q is the flux rate, L the length of the column and AH the hydraulic head difference
across the column (Fig. 3.3).
Hydraulic conductivity, K, is a measure of fluid flow through porous media. It is

a firnction of both the fluid and soil properties.

K: We
,1:

(5.3)

64

where k is the permeability of the soil, p is the density of the fluid, 11 is the viscosity of
the fluid and g is acceleration due to gravity. In a two— or three-phase system,
permeability is also considered to be a firnction of the fluid phase saturations (Section 2.3;
Fig. 2.5). The relative permeability, k,, is the permeability scaled by the saturated
permeability, km, which in these experiments is the saturated permeability of water prior

to surfactant enhancement.

K k
k = — = — 5.4
' Ksar km! ( )

5.3 Results and Discussion

Prior to measuring the relative permeability of the aqueous phase at $0 2 1 - Sm, a
range of Sm values was measured for the natural unconsolidated soil studied.
Experiments were then conducted to measure the relative permeability of the aqueous
phase over a range of aqueous phase saturations in this region of discontinuous
nonwetting fluid. Additional experiments were also conducted to measure the relative
permeability of the aqueous phase in the presence of both trapped residual NAPL and a
trapped residual gaseous phase. In these experiments, each data point represents a
separate column experiment, assuming that the saturation of the residual trapped NAPL
and the trapped residual gaseous phase was uniform. A third experiment was conducted
to measure the increase in the relative permeability of the aqueous phase in a single
column study as the saturation of the trapped residual NAPL was decreased through the
dissolution of the NAPL into a surfactant enhanced aqueous phase. The data was then
used to evaluate to existing models for predicting the relative permeability - saturation

relationship.

65

5.3.1 Trapped Residual NAPL

A range of aqueous phase at $0 2 1 - Sm, were measured by changing the ratio of
the capillary to viscous forces, as represented by the capillary number, Ne (Eqn. 2.5) as
presented in Table 5.1. Past researchers have estimated that No, needs to be > 10'4 to
minimize residual entrapment or mobilize trapped residual NAPL (Wilson et al. 1990;
Chatzis and Morrow 1984; Larson et al. 1981). Unless the IFT is reduced, a capillary
number this high would requires nondarcian flow rates which are almost never observed
in nonindurated rocks and granular material under natural conditions (Freeze and Cherry
1979). The upper limit of Ne within the range of Darcian flow can be estimated by Eqn.
2.5 and Eqn. 3.4 assuming 1< Re <10.

< Rey2

Ne-
pda

 

(5.5)

In Table 5.1, a bulk density of 1.81 i 0.03 (1.5%) and a porosity of 0.39 i 0.01 (3.4%)
indicate column packing and initial aqueous phase saturation reproducibility. For the
total range of Nc, Sm ranged between 131-36.5%, excluding one unexplainable
measurement of 5.1%. The range of Sm is reduced to 136-32.6% if the upper limit of
Darcian flow is assumed to be at Re =1. This range is a reflection of the variable ganglia
sizes found in Ottawa sand (Powers et al. 1992). Table 2.1 reviews past research in
trapped residual NAPL which indicates that a range of trapped residual NAPL can exists
in sands. A similar review is presented in Mercer and Cohen (1990).

In these experiments, and in past research, the size and the shape of the trapped
residual NAPL is assumed to be uniform. However, trapped residual NAPL has been
cast in place in sandstone, glass beads and sand, illustrating the wide range of ganglia I
' sizes and shapes, by polymerizing styrene under high temperature and pressure (Chatzis
et al. 1983; erson et al. 1990; Powers et al. 1992). A single isolated drop in a pore
space can exist within the same region as elongated multiple pore ganglia or ganglia

branching throughout a cluster. These blob casts were made from column experiments

where the size of the column may have a limiting influence on the size and shape of the
larger blobs which may not occur in the field.

Hunt et al. (1988) conducted a review of previous work in petroleum engineering
and fluid mechanics in NAPL emplacement and mobilization from a different viewpoint.
In this review, equations were presented to provide an order of magnitude estimate of the
maximum stable length of residual NAPL blobs, also referred to as ganglia, in both the
saturated and unsaturated region. The following text summarizes their findings.

In the saturated region the maximum vertical length of a ganglia, meax can be
estimated by the Hobson Equation (Berg 1975):

20”,,

L
r. glApl

v,max

 

Ill

(5.6)

In this equation r, is the radius of the smallest constriction of a pore (throat radius), g the
gravitational constant, and the interfacial tension between the NAPL and the aqueous
phase and Ap the density difference. For horizontal flow, the maximum ganglia length,
Lh,max> can be estimated by :

0,0216 203... (5.7)

Lt. .
.Uav’} Paglrr

 

 

Ill
Ill

,max

where k is the permeability of the soil, pa the viscosity of the aqueous phase, v velocity, p
a the density of the aqueous phase and i the hydraulic gradient. In the vadose zone, the

stable ganglia length, Lv’max, can be estimated as:

2
LW 5 ”"8 (5.8)
r. p. g

 

where ong is the interfacial tension between the NAPL and the gaseous phase and p" is
the density of the NAPL.

The radius of the throat (r,) may be estimated fi'om a rhombohedral packing of a
spherical media with a diameter d (Berg 1975): '

67

r, z 0. 077d (5.9)

Using the soil and fluid properties in Table 5.2, these equations were used to
provide an order of magnitude estimate for the maximum size of a stable trapped residual
ganglia of NAPL, hence predicting the potential for a large range of ganglia sizes and
shapes Since Nb > 0.00667 (Sec. 2.2; Fig. 2.3), No can be used to estimate the trapped
residual volume of NAPL prior to surfactant injection. Typical flow rates in sand and
gravel aquifers range from 10m/y to 100 m/y (3 x 10'5 cm/s to 3 x 10‘4 cm/s) (Mackay et
al. 1985), resulting in 6 x 10'9 3 Ne _<_ 6 x 10's. The maximum vertical stable ganglia
length in the saturated region is estimated at 112 cm., whereas the maximum horizontal
stable ganglia length is estimated between 90 - 900 m. For the vadose zone, the
maximum stable vertical ganglia length is estimated at 18 cm. It is reasonable for
hydrological cycles to drop the water table greater than 18 cm., resulting in such an
elongated ganglia (i.e., in Reichmuth 1984 the water table fluctuated 142 cm. between
September and February due to seasonal recharge/discharge). When the water table rises,
these ganglia will be captured in the saturated region. Calculations support that these
ganglia are well below stable vertical lengths for the saturated region. As a result, a rising »
water table will not mobilize these ganglia.

When estimating the maximum horizontal stable length in the saturated region as a
function of gradients, a 1% gradient (i = 0.01) can trap a stable ganglia 27 m long
whereas a 10% gradient can maintain a stable ganglia length of 3 m. This explains in part
why pump and treat technology has limited use in the remediation of LNAPL. When a
drawdone cone develops in the are of a pumping well, a portion of the pool of LNAPL
will migrate under the influence of the locally steep gradients. These calculations predict
that stable ganglia can remain trapped over a large region as opposed to reaching the well
where the NAPL can be removed. As stated previously, in the saturated region it is

estimated that 10-50% of the pore space is saturated with trapped residual NAPL ganglia.

68

As stated, these estimates only provide an order of magnitude estimate for the
maximum stable ganglia lengths. Much smaller lengths were found in the polymerized
styrene blobs, which at the most extended over a few pore spaces. The purpose of
evaluating the stable ganglia length was to suggest that trapped residual NAPL
saturations in the field may be significantly higher locally, resulting in significantly higher
reductions in the relative permeability of the aqueous phase. Column experiments
provide a better understanding of the saturation-permeability relationship in the range
measured, but it is not clear it adequately describes the in-situ environment without

further exploring 2-D and 3-D situations.

5.3.2 Reduced Permeability Of the Aqueous Phase Due to Trapped Residual Phases

In order to simulate conditions in the upper region of the capillary fiinge, water
saturated columns were flushed with air followed by water to trapped residual air phase.
The trapped residual immiscible phase reduced the relative aqueous phase conductivity to
0.63 i 0.04 (Fig. 5.1). Morrow et al. (1988) reported that the relative permeabilities for
bead packs were largely independent of establishing the residual saturation by
mobilization or entrapment and the phase of the trapped residual (e.g., gaseous phase or
oil phase). Therefore, the reduction in relative permeability may also represent a region
within the capillary fringe or below the water table where trapped residual NAPL only is
trapped. Trapped residual air saturations were measured between 11.27-15. 14%
(average 13.6; standard deviation 4.5), which is consistent with the saturation of NAPL
measured previously. The columns were then flooded with NAPL, followed by a water
flush to establish a trapped residual NAPL phase in addition to the trapped residual air
phase. The trapped residual air is largely in the larger pore spaces prior to the NAPL
infiltrating the column. Trapped residual NAPL saturations were estimated based on the
assumption that the trapped residual air phase remained within the soil matrix. Between

4-30% (average 17.43; standard deviation 8.5) of the pore space was occupied by a

69

trapped residual NAPL, reducing the relative permeability of the aqueous phase to 0.32 i
.02 (:1: 6%).

Experiments were also conducted with only NAPL trapped as a residual
immiscible fluid. In these experiments, the aqueous phase saturation was reduced to
69.6-84.5% (average 80.1; standard deviation 5.1) . As a result, the relative permeability
of the aqueous phase was reduced to 0.52-0.74 (average 0.6; standard deviation 0.08).
These results are comparable to a reduction in the relative permeability of the aqueous

phase to 0.6 when a trapped residual air phase was established in the water saturated
I column, although the range of saturation for the trapped residual NAPL only is slightly

larger.

5.3.3 Aqueous Phase Permeability During Surfactant Enhanced Dissolution of NAPL

A trapped residual NAPL phase saturation of 17.9% was established in a soil
column followed by continuous injection of 5% solution of Witconol NP-150 for 14
days. The relative permeability of the aqueous phase was measured as a function of pore
volumes by monitoring the head loss across the soil column using a water manometer.
The results indicate that the trapped residual NAPL reduced the relative permeability of
the aqueous phase to 0.52. The initial surfactant injection firrther reduced the relative
permeability of the aqueous phase to 0.35 which indicates that the viscosity and density of
the surfactant is different that deionized water used to meaure the conductivity
previously. After 14 days of continuous pumping, the relative permeability of the
aqueous phase increased to 0.58. A final flush of 4 pore volumes water further increased
the permeability to 0.92. Surfactant injection without trapped residual NAPL reduces the
aqueous phase relative permeability to 0.78, which again indicates a difference in the
viscosity and density of the surfactant compared to the deionized water used to measure
the saturated conductivity. The final flush of water may not have completely removed all

of the surfactant which would result in a lower aqueous permeability as well as a lower

70

weight measurement for the final determination of NAPL removal, since the density of
surfactant is less than the density of water (Table 5.4). Weight measurements estimated a
complete removal of the NAPL although further pulsed pumping of surfactant indicated
filrther NAPL removal.

If the remediation design was based on pumping surfactant into a contaminated
region without consideration for the reduced flow within the region of reduced
conductivity, the estimated flow into the region of contamination would have been in
error since flow into the contaminated region is reduced to approximately 52% by the
method of Wheatfield and Winterberg. Toward the end of the treatment, the flow had
increased to 73%. This range of reduced flow, although estimated on the basis of an
idealized homogeneous uniform flow, suggests that pumping strategies should account
for reduced permeability that will increase over time in order to develop an economical
remediation technology.

This experiment was conducted in a naturally occuring unconsolidated soil as
opposed to consolidated sand stone or glass beads common research conducted for
petroleum reservoirs. It is also unique in that the data is collected continuously in a single
column as opposed to collected data over a series of column experiments where the
distribution of the trapped residual non-wetting phase or soil packing may vary. The data
follows the trend of the data collected in the previous sets of experiments, indicating that
the final combined graph can be used to predict the permeability regions near the capillary
fiinge for the aquifer material studied.

5.3.4 Comparing Data in Discontinuous Region with Existing Fitting Parameters
Figure 5.2 contains data derived from a series of experiments where the range of
trapped residual NAPL saturation measured in an unconsolidated porous media was

varied by changing the ratio of the viscous and capillary forces. In addition, data is

71

included from an experiment where the trapped residual NAPL was reduced through
dissolution into a surfactant enhanced aqueous phase.

A discussion of the relative permeability-saturation curves is presented in Sec. 2.3
which describes the 2-phase relationship in detail. Empirical equations have been
developed based on experimentally derived coefficients or fitting parameters A, 01', n’ and
m for both pressure-saturation relationships and relative permeability-saturation
relationships. Corey (1986) defines these relationships as a function A and effective

saturation (Se) of the aqueous phase.

11
S, =Sl—“__S—SM-=(%) (5.10)
km = Sf?" (5.11)

where S0 is the saturation of the aqueous phase, Sm the residual or irreducible saturation
of the aqueoUs phase, 11,, the displacement pressure head, It pressure head, and km the
relative permeability of the aqueous phase. Van Genuchten (1980) defines these

relationships using 0', n' and m:

 

_&—&~ 1 ”
S" l-Sm :(1+(ah,)") (5'12)
km =S}{1—[1—S,‘i5)]"} (5.13)

where he is the capillary pressure. head. It is usually assumed that m = 1-1/n'. Both 11'
and 1 are measures related to the pore size distribution index such that n' = 2. + 1 (Powers
et al 1992; van Genuchten, 1980).

Table 5.1 summarizes values of A, 01', n' and m reported in the literature for
Ottawa sand. Soeryantono et al. (1994) used the same soil analyzed throughout this
research. Powers et al. (1992) used a 20-30 mesh Ottawa sand purchased from Fischer

72

Chemical Co. In both works, the fitting parameters were derived from data measuring
the pressure-saturation relationship.

In Figures 5.4-5.8, the fitting parameters for both Corey's and van Genuchten‘s
relationships were used to estimate the relative permeability saturation relationship in the
region where both the wetting and non-wetting fluids are continuous. The curves were
then extended into the region where the non-wetting fluid is trapped over a range of
trapped residual non-wetting saturations using the same equations and fitting parameters.
The irreducible aqueous phase saturation (Sm = 0.093) was measured by Soeryantono et
al. (1994) from the same Ottawa sand used in this research. Figures 5.4-5.6 use the
fitting pararnetes measured by Soeryantono et al. (1994) over the range of possible
values (e.g., A = 2.32 :t 0.02; n' = 4.27 i 2.72) Powers et al. (1992) values for imbibition
and drainage are presented in Figures 5.7-5.8. In all cases except the use of the van
Genuchten parameters in Figure 5.4, the extended curves predicts the trend in the lower
region of the data to some degree. However, none of the curves describe the data in the
upper region. The curves do predict that a small change in the saturation of the trapped
residual non-wetting phase results in a larger change in relative aqueous phase
permeability compared to the regions where the nonwetting phase was continuous. The
results indicate that the existing fitting parameters do not adequately predict the
permeability-saturation relationship in the region of interest. Therefore, the raw data may

need to be used directly in design and modeling pumping strategies in this research.

5 .4 Summary

Experiments were conducted to measure the relationship between the relative
permeability of the aqueous phase and the aqueous phase saturation in a two-phase
system where the nonwetting phase is discontinuous. Over a range of trapped residual
nonwetting saturations between 11.27-15.14% the relative aqueous phase permeability

was reduced between 0.6k,. to 0.8/tr. The range of trapped residual nonwetting phase

73

saturation was established by varying the ratio of the viscous to capillary forces. Similar
studies have been conducted by petroleum engineers in sandstone and glass beads, but to
the knowledge of the author, this study is unique in that it was conducted for a naturally
occuring unconsolidated soil. The study was extended to measure the relative
permeability of the aqueous phase with both a trapped residual NAPL and a trapped
residual air phase established. Using the columns from the previous set of experiments
where trapped residual air saturations were measured between 11.27-15.14%, an
additional trapped residual NAPL was established between 4-30% saturation. The
combined trapped residuals reduced the aqueous phase to approximately 0.32k,..

In the research conducted by the petroleum engineers, this type of experiment
provided a measure of the relative permeability of the aqueous phase in the region of
trapped residual NAPL but also provided a measure of pressure gradients or reduction of
interfacial tension needed to decrease trapping during water flooding or increase
mobilization of the trapped residual NAPL. The pressure gradients needed to decrease
the amount of trapping in groundwater aquifers are unrealistic. Releasing trapped
residual NAPL by reducing the interfacial tension threatens to spread the contaminant
further through the aquifer. This research considers the use of a surfactant to enhance the
solubility of the trapped residual NAPL into the aqueous phase. While many in-situ
remediation requires a measure of the reduced permeability in the contaminated region,
the dissolution of the trapped residual NAPL requires a measure of the increasing
aqueous phase permeability over time. An initial design based on promoting both the
flow through a contaminated region of reduced aqueous phase permeability and the
contact time needed for optimizing the solubility of the NAPL may need to be modified as
the flow paths change due to the reduced saturation of the trapped residual NAPL and a
concurrent increase in the relative permeability of the aqueous phase. Under constant
head boundary conditions, the increase in permeability will result in an increased aqueous

flow into the contaminated region. Under constant flux boundary conditions, the

74

increased permeability will result in an increase contact time of the surfactant enhanced
aqueous phase (Wilson et al. 1990). An additional column experiment was conducted to
measure this relationship. In addition to providing the design parameters needed for this
research, this experiment is a unique approach to measuring the permeability-saturation
relationship in the region considered.

One may argue that in an actual sandy aquifer targeted for remediation, the
reduced relative permability in the contaminated region due to trapped residual NAPL
may be minor when considering local heterogeneities, such as silt or clay strata, which
may reduce the permeability by a factor of 1000 to 10 million (Mackay and Cherry 1989).
In groundwater aquifers, capillary forces are generally stronger than the combined
viscous or buoyance forces. As a result, NAPL will be trapped in the larger pores of a
sandy region as opposed to migrating or trapping in the smaller pores in the less
permeable regions. The clay and silt layers may provide regions within the aquifer which
locally represent an confined aquifer promoting flow into a contaminated lens of soil,
while causing unpredicted surfactant by-passing in another area. The size and shape of
the contaminated region relative to the size and shape of the region of clay or silt needs
to be considered when characterizing the region and evaluating the use of hydraulic

controls to efficiently deliver, recover and promote surfactant enhanced dissolution.

75

 

 

0.9 --
0.8 *-
0.7 -~ g
*4
_ O
0'6 , 9 Trapped residual air ..
O

O Trapped residual air and oil

 

 

 

RELATIVE PERMEABILITY
.6 .c .<= .o
N U A M
i l l i

O

0

<6)

0

.O
a—s
1
l

 

 

0 i i i i i i i i i l
0 10 20 3O 40 50 6O 70 80 90 100
AQUEOUS PHASE SATURATION (%)

 

 

 

Figure 5.1. Measured values of relative permeability and saturation of the aqueous
phase with trapped residual air and trapped residual oil.

76

 

 

 

 

 

 

 

l ....
0.8 ~
g 4
k1 OnesidunlExp. :
h 06 .0. ODissohm'onExp. .?
' 4
E a
«so
3 83:
a 0.4 r o
H
8
S .
At
‘8
o 0.2 ~~
rd
:3
O
<
O 1 i i i 4
0 0.2 0.4 0.6 0.8 l
AQUEOUS PHASE SATURATION

 

 

 

Figure 5.2. Aqueous phase relative permeability as residual NAPL is reduced
through surfactant enhanced dissolution.

 

 

 

A 3.5-

a l

r. 3.45 "

E 3. .

0

>

g 3'35 ‘ W00

3 3.3 . ° . - - -
0 0.2 0.4 0.6 0.8 1

Relative Permeability

 

 

 

Figure 5.3. Error associated with relative permeability measurements.

.63 u ..H .8." u «...H... v «.8... ..H
5:233. 3.2..— 3323 ..H—.28.. a 5; 3a.... 282:3 2: .3 35:83.9. =2.§5a?bm=aao::2_
9529.. 2: .8.— ..85556 S; can .380 .3 335.8.— ..8595 8 8a.. aura—.50 in flawfi

 

 

77

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

zofigé mam: 38:3 20:55 mafia 88:3
m wd ed v.0 «.9 Q ~ wd 0.6 #6 N6 o

w n H ...H ........ n o T \ ....H ....... n c
m m
m m
., do m ., 3 m
S S
W O .d
S 9..... M
-, to m o... -. to m
w M. m.
32.3: o u w . Spas: 0 u
3 -. H; m 9 ..H... .- 2H m
8535 0 M. Q Sneeze O m...“
5209185 ....................... m 0 F5385 ....................... M
638605» 11 we W 62860.! lac W
m m

l — l m

 

 

 

 

 

.5... u ..H .8." u 2:5: V «.85 ...H
5:23am 85... 252:3 1328.. a 5m? 82:. 232...: 2: .3 3:28:22 =o:a..3a?b=3ao::on
3:29.. 2: .8.— ..ofioasmv =5» .25 .380 ..o 3.3—:8 :32an 8 3a: wctanEoU .m.m 9..—arm

 

 

78

 

zofifispé mafia 88:3 zoPfiSSm mafia 8953
H 3 9o EH 3 o _ no 3 EH 3 o
1.-» » o » » »1..».. + o

P h F
u «

 

 
    

 

 

0
0......
w .96 2&5. O
Q
.~

,9
.1 ed .3 r 9o
O 53335 O
....................... 3.60.385
% wd fizuauoglll .1 wd

 

 

 

 

 

 

ALI'IIHVEIWHEH ElAILV’IEl‘d ElSVHd snoanbv
wnavamad HALLV’IERI EIS'v'Hd snoanbv

 

 

 

 

 

 

 

.83 n ..H .3." u 3...... v 3...: 2
8:82.23 82:. 282:8 8:28.. a _:_3 82..— 8853 2: .8 3:88:29. 8:588?b=33=:2_
3:50.. 2: .8.— .822890 E; .28 .380 .8 8888:3188». 8 52. 2:89.80 9m 082m

 

 

79

 

zoESHPSHm mafia 8089. 205233 ESE 3859.
H no we ...H. 3 o H 3 3 to S o

p
u

«L-
T4»-
‘.
#-
0

4p
p—o—
o
‘V"

b
q u

1

l
N.
O

l

l
N.
O

l
l
V.
o

 

 

l
l
‘5!
o

 

1
l
°°.
o

 

 

 

 

 

 

umavawuad HALLV'IEDI EISVHd snoanbv
umavamad HALLV'Izm ElSVHd snoanbv

 

 

 

 

 

 

 

 

.93. u ..H .3." u 2:5. » n8... 2
8:988» 8228 282:8 8:28.. a .:_.$ 3.28 2.82.8 2: .8 3:28:22 =3:E=8?b=338.8:
3:50.. 2: 8.. 282.2289 28> .28 .880 .8 888.8.— ..32828 8 «22. 8:22.80 Em 9.82m

 

 

80

 

zoFSSEm was: gonzo... 885% as: 8859.
H 3 3. ...H. 3 o H no 9o 8 S o

 

 

_ O F _ L b _ o
H q H H H

l

l
N.
o

l

l
“l
o

 

 

 

O
O
% ..me 2E5 0
a.
..

O
1 9o ...... 1 9o
O 8:230 O
....................... in
1r ”.0 52350:!l .... w.o

 

 

 

 

 

 

mmavawaad ElAIlV’IERI EISVHd snoanbv
umavawuad EIAIlV'IEI‘d aSVHd snoanbv

 

 

 

 

 

 

 

 

A": u ..H .21... u 3...... v «.25 2
8:988... 0228 88:8 8:38.. a :23 8228 88:8 2: 8 2:88:29. 8:8:8»-§_E8E8:
3:28.. 2: .8.— .828280 28> :8 8.80 .8 22:88.— _8:_:Eo 3 :8: wing—80 .w.m 8.82.:

 

 

81

 

 

 

 

 

 

 

 

 

 

 

 

 

20:55 as: 2883 2053593 mmsa 88:9.
H no 8 to No o H no we ...o No o
.r » H ........ » o H H » H ......... H o

V V
O O
m m

1: NO m 1: No m
w. m

0

, 3H N» Am to m»
m ......_ .
v w
u w .3385 O u

H- 8 m Q .3 .- 9o m
mm ” 89:08.0 0 mm
m 0 8.80.385 ....................... m

1: ”.0 W fiEoBBOEll i ”.0 W
H". m

» w M

- H .- H

 

 

 

 

 

 

 

 

82

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

Nc Error Sm Error Relative Darcy Re pd n
Error(%) velocity’
(ml/min)
2.46E-04 1.231504 0.365 0.008 0.021 0.635 2.678 ND 1.789 0.402
1.64E-03 8.22E-04 0.365 0.008 0.021 4.232 17.859 ND 1.801 0.401
2.45E-04 1.231304 0.131 0.003 0.026 0.635 2.678 ND 1.766 0.403
1.65E-03 8.26E-04 0.204 0.005 0.023 4.232 17.859 ND 1.792 0.399
3.30E-06 20413-06 0.165 0.004 0.024 0.008 0.035 D 1.818 0.389
6.75E-04 3.37E-04 0.154 0.004 0.025 1.693 7.143 ND 1.822 0.390
3. 1913-06 1.98E-06 0.172 0.004 0.024 0.008 0.035 D 1.792 0.402
2.43E-05 1.22E-05 0.247 0.005 0.022 0.063 0.267 D 1.801 0.405
2.68E-05 1.35E-05 0.326 0.008 0.025 0.063 0.267 D 1.833 0.364
2.29E-03 1.15E-03 0.051 0.002 0.044 5.925 25.003 ND 1.770 0.402
7.09E-05 3.55E-05 0.201 0.005 0.023 0.169 0.713 D 1.835 0.371
1.06E-04 5.291305 0.136 0.004 0.026 0.254 1.071 D 1.854 0.373
6.8013-05 3.40E-05 0.179 0.004 0.024 0.169 0.713 D 1.839 0.387
2.91E-06 1.461506 0.145 0.004 0.030 0.169 0.713 D 1.778 0.389
1.807 0.391 avg
D=darcian 0.027 0.013 std
ND=nondarcian ‘ flushing velocity 1.520 3.409 cv(%)

 

 

Table 5.1. Capillary number vs. saturation.

 

83

 

 

 

 

 

 

 

 

 

 

 

 

 

 

Parameter Value Method
d 0.05 cm d50; sieve analysis
Pn 0.748 :1: 0.0003 g/cm3 measured 25°C
Pa 0.998 :1: 0.0004 g/cm3 measured 25°C
ps 0.995 :1: 0.0032 g/cm3 measured 25°C
°na 52.8 :t 0.2 dynes/cm 25°C; Johnson and Dettre 1966 .
Orig 24.9 :t 0.2 dynes/cm 25°C; Johnson and Dettre 1966
°ns 3.73 :t 0.09 dynes/cm 25°C; Pendant Drop Method
pa 0.01 poise 20°C; Potter and Wiggert 1991
k 10*5 cm2 Freeze and Cheny 1979
Subscripts Aqueous phase (deionized water)

NAPL (dodecane)
Gaseous phase (air)

Surfactant (5% W i tconol NP- 150)

 

Table 5.2. Soil and fluid properties.

 

84

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

Minimum Maximum Wetting phase NAPL

Flow Rate 3 x 10'5 cm/s 3 x 10‘4 cm/s water dodecane

Nb 2.9 x 10‘5 n/a water dodecane

Nc 6 x 10-9 6 x 10-8 water dodecane

Lv,max saturated 112 cm n/a water dodecane

Lh,max saturated (1) 900 m 90 m water dodecane

Lh,max saturated (2) 27 m n/a water dodecane

Lh,max saturated (3) 3 m n/a water dodecane

Luna: unsaturated 18 cm n/a water dodecane

Lh,max saturated (1) 63 m 6 m 5% Witconol NP-150 dodecane

Nb 0.04 Na 5% Witconol NP-lSO dodecane

Nc 8 x 10'8 8 x 10‘7 5% Witconol NP-150 dodecane
(1)estimated from velocity

 

(2)estimated from gradient i = 0.01

 

 

(”estimated from gradient 1 = 0.1

 

Table 5.3. Order of magnitude estimate for ganglia size

 

85

 

 

 

 

 

 

 

 

 

 

Parameter Value Source

a'i 0.222 Powers et al. 1992
“'d 0.094 Powers et al. 1992
n'i 3.58 Powers et al. 1992
n'd 7.42 Powers et al. 1992

n' 4.27 :l: 2.72 Soeryantono et al. 1994
11' 2.30 Powers et al. 1992
id 3.46 Powers et al. 1992

3. 2.32 :t 0.02 Soeryantono et al. 1994

 

 

 

 

Table 5.4. Reported values of fitting parameters for white Ottawa sand used in
relative permeability-saturation and capillary pressure-saturation relationships in a
2-phase system (i = imbibition; d = drainage).

Chapter 6
Numerical Modeling

6.1 Introduction
Leaking underground storage tanks, illegal disposal of waste and pipeline ruptures

results in subsurface spills of nonaqueous phase liquids (NAPLs) which contaminate the
groundwater. As NAPL migrates through the soil matrix under the influence of capillary,
viscous and buoyancy forces, a residual saturation of NAPL will remain in the pore
spaces. In the unsaturated region, residual NAPL has been traditionally viewed as a
continuous film between the aqueous and gaseous phase present in the pore spaces.
However, recent flow visualization experiments conducted in micromodels suggest that
nonspreading NAPLs may actually coalesce as discontinuous ganglia on the aqueous
phase interface (Wilson 1994). Larger spills which are not completely reduced as residual
in the vadose zone will pool above the largely water saturated pores of the capillary fringe
until sufficient pressure is developed to drain the aqueous phase from the pores. NAPL
less dense than water (LNAPL) such as natural and refined hydrocarbons (Knox and
Sabitini 1991) will continue to pool while slightly depressing the capillary fringe.
Seasonal watertable fluctuations or plume migration due to natural gradients can cause
this pool of NAPL to become discontinuous and immobile as a trapped residual phase.
NAPLs more dense than water (DNAPL) such as chlorinated organics and solvents
(Knox and Sabitini 1991) will migrate through the capillary fiinge. The migration of
DNAPL will continue through the saturated region until it is completely exhausted as a
discontinuous and immobile trapped residual phase or pooled above less permeable strata

in the aquifer. Despite the reduced saturation of residual NAPL remaining in the pore

86

87

spaces, it creates a long term source of pollution partitioning slowly into the aqueous and
vapor phases (Schwille 1988; Mackay et al. 1985).

Trapped residual NAPL is considered one of the major limitations to conventional
pump-and-treat technology (Schmelling 1992; Hall 1989). The chemical enhancement of
pump-and-treat methods with surfactants and cosolvents has been studied to improve the
technology (i.e. Ellis et al. 1985; Nash 1988; Abdul et al. 1992; Clarke et al. 1992;
Pennell et al. 1993; Fountain 1991). In a 1992 report , the Environmental Protection
Agency identified key areas of concern for chemical enhancement of pump-and-treat
remediation which include as the delivery of the reactive agent to areas of the aquifer
where it is needed and the removal of the reactive agent from the subsurface (Palmer and
Fish 1992).

NAPL blocks or reduces the flow of the aqueous phase into the contaminated
region. This in part explains why the limited number of field applications to date of in-
situ remediation technologies for the treatment of NAPL have not achieved the high
efficiency suggested by 1-D column experiments (Ellis et al. 1984; Nash and Traver 1984;
Nash 1985). The columns impose an artificial boundary which controls fluid movement
between injection and recovery points. However, in the field flow is free to by-pass the
NAPL contaminated region due to the resistance caused by this immiscible phase within
the soil pores. This resistance can be decreased if the NAPL saturation is reduced
through pumping or natural plume migration. The remaining discontinuous residual
saturation is far more difficult to remove, trapped by capillary forces stronger than the
combined opposing viscous and buoyant forces. In the saturated region, residual NAPL
occupies between 10-50% of the pore space (Wilson and Conrad 1984). Hence, the pore
blockage resisting flow is still significant. The problems caused by blockage are
accentuated in the capillary fiinge, where a residual gaseous phase is also trapped within

the soil pores, fiirther reducing the aqueous phase permeability.

88

Field data and column studies have shown that the extent of interphase mass
transfer from the NAPL to the aqueous phase are limited by chemical or physical
processes (Power et al. 1992; Mackay et al. 1985; Mercer and Cohen 1990; Miller et al.
1990). Powers et al. (1992) suggests that these limitations are due to: 1) rate-limited
interphase mass transfer (nonequilibrium); 2) physical bypassing of the mobile aqueous
phase around contaminated regions due to relative permeability effects or aquifer
heterogeneities; and 3) dependence of equilibrium solubilities on NAPL composition.
Equilibrium mass transfer assumes that the maximum solubility of NAPL into the
surfactant enhanced groundwater will occur instantly. Rate-limited interphase mass
transfer may result in less than predicted removal rates if the time needed to promote the
maximum solubility of NAPL into the surfactant is not achieved between the injection and
recovery wells. In contrast, continuing to pump surfactant which has reached the
maximum solubility of NAPL through a polluted region is justifiably as inefficient. Less
than optimal interphase mass transfer may also be a result of in-situ physical limitation.
In the complex network of pores spaces within the soil, trapped residual NAPL may be
isolated as a single blob or branched through several pores spaces. As a result, the
solubility of NAPL into the surfactant enhanced groundwater may be limited at the pore
scale by dead end pore spaces or groups of blob blocking flow , restricting flow or
reducing the contact area for mass transfer.

As the surfactant enhanced aqueous phase reduces the NAPL saturation through
dissolution, the permeability will increase within the region. Under constant head
boundary conditions, the increase in permeability will result in an increased aqueous phase
flow into the contaminated region. Under constant flux boundary conditions, the
increased permeability will result in an increase contact time of the surfactant enhanced
aqueous phase within the contaminated region (Wilson et a1. 1990). As a result, the
economical and efficient use of surfactants to enhance the dissolution of NAPL requires

coupling the accurate prediction of rate-limited mass transfer as well as the relative

89

permeability of the continuous aqueous phase over time. This research will incorporate
previous 1-D column experiments designed to investigate these to phenomenon in a 2—D

study.

6.2 Objective

The overall objective of the research will be to evaluate hydraulic controls to
promote the efficient use of surfactants to enhance the dissolution of trapped residual
NAPL near the capillary fiinge. The specific objectives are 1) to develop an implicit
numerical transport code to optimize the pumping rates in a nondimensional l-D study; 2)
to evaluate the use of the USGS model SUTRA (Saturated-Unsaturated Transport
Model; Voss 1984) to model a vertical cross section of an aquifer which includes the
vadose and saturated regions; 3) use SUTRA to predict the capture zone and contact
times established by hydraulic controls (i.e. pumping rates and well placement) in a region
near the capillary fiinge where the relative permeability of the aqueous phase is reduced
due to a trapped residual NAPL; and 4) to evaluate the effects of increased permeability
due to dissolution within the region on the design of the system.

Column studies have been conducted to measure the kinetics of rate-limited mass
transfer and reduced permeability due to trapped residual NAPL near the capillary fringe.
In these experiments, a 5% solution of a nonionic surfactant, Witconol NP-150
(ethoxylated nonylphenol) was injected into Ottawa sand with trapped residual dodecane
(Chapter 4). The column studies have shown the presence of mass transfer limitations,
but the kinetics are not prohibitively slow. A contact time of 12 hr. is needed to achieve
the maximum dissolution in the porous matrix of NAPL into a 5% solution of Witconol
NP-150. Trapped residual NAPL reduces the aqueous phase permeability by
approximately 30%. However, at the capillary fiinge, the gaseous phase can also be
trapped as a discontinuous phase, firrther reducing the aqueous phase permeability by

approximately 50%. The 1-D column studies also measured changes in the mass transfer

over time as the contact area of the NAPL changed in response to dissolution. The
column studies also measured the increase in relative permeability as the trapped residual
NAPL saturation was reduced by the surfactant. This research will use the data from the
1-D experiments to continue the study of surfactant enhanced aquifer remediation in a

2-D environment.

6.3 1-D Model Development

The numerical evaluation will consider a single component NAPL present at
trapped residual saturations. The immobilized ganglia are uniformly distributed in a
saturated, homogeneous isotropic aquifer material. NAPL transport into the aqueous
phase can be predicted by the 1-D advection dispersion reaction (ADR) equation for a

homogeneous medium with steady flow in the x—direction:

5C 32C 5C
—-=D — —- J 6.1
51 6x2 v§x+ ( )

 

In the ADR equation, C is the concentration of the solute, t is time, D is the
dispersion coefficient, v is velocity, x is distance and J describes the nonadvective flux or
dissolution of NAPL into the aqueous phase. Assuming that mass transfer is adequately
described by a linear driving force model, J can be modeled as the product of a lumped
mass transfer coefficient, k,, the difference in the equilibrium concentration of the solute

in the aqueous phase, CS, and the concentration of the solute present in the aqueous
Phase C:
J = k, (C, - C) (6.2)

The reduction of the immobilized ganglia is estimated from the mass balance:

ac”
a:

 

= -k,(C, — c) (6.3)

where C N is the concentration of the NAPL.

91

An implicit numerical transport model (Chapra and Canale 1988; Wang and
Anderson 1982) was developed using centered finite divided difference for the spatial
terms and block centered nodes, resulting in a tridiagonal matrix. The concentration at a

node is weighted over time by a) .

 

 

CviH-l _ Ci, : 0D Cull-:11 _ 2Cil+l + Cilj-ll + (1 _ (0)0 le+1 _ 2C: + (Iii—1
Ar sz A x2
1+1 _ 7+1 1 _ I
— wvgliL—gl— - (1 — co)v——C’+1 C” (6.4)
2Ax ZAX

+ wk,(C. — Cf“) +(1- w )k, (C. -C.-’)

The subscript refers to the x-location whereas the superscript references time. The
numerical code was written in FORTRAN. Simulation were performed on an IBM
compatible 386/486 computer using a 32-bit compiler.

The model assumes that the NAPL is a nonwetting fluid. Consequently, the
wetting fluid, in this case the aqueous phase, acts as a barrier to direct contact between
NAPL and soil particles, strongly inhibiting the mass exchange which may occur between
the NAPL and the soil or solid phase. Prior to surfactant treatment, the trapped residual
NAPL is assumed to be uniformly distributed throughout the matrix. Over time, the
distribution will become less uniform as a result of the stronger driving force of the
nonadvective flux term at the influent end. The aqueous phase solubility of the NAPL is
assumed to be low enough that the solute concentration does not affect the density of the
contaminated aqueous phase.

To minimize numerical dispersion, time was discretized to maintain the Courant
(Cr) number less than or equal to one: '

Jfl
Ax

Cr 3 l (6.5)

and the size of the grid met the criteria for the grid Péchlet number:

92

Ax 3 4a (6.6)

where or is the dispersivity of the solute (Anderson and Woessner 1992).

An analytical solution (van Genuchten and Alves 1982) was used to verify the
code during early time when the saturation of the trapped residual NAPL is relatively
uniform and constant. The analytical solution for a semi-infinite column exists with a
first order decay term and a zero order production term for the following boundary

conditions.

C(x,0) = C,

C(O, r) = C, 0 < r s to (6.7)

é’C(oo,t)
at

For the system under consideration, C0 , the initial concentration of NAPL in the injected
surfactant, and C,- , the concentration of NAPL in the aqueous phase present in the soil
matrix, are both zero. The solution assumes that the dissolution of NAPL into the
aqueous phase begins at time to for all elements and that the surfactant concentration of
the aqueous phase is constant. The complete analytical solution is in Appendix A.

The 1-D ADR equation can be transformed to a nondimensional form by defining

the following dimensionless terms.

 

 

X:i Daz-ki
L v
C=£ Fe:£ (6.8)
C, D
t*=’—"
L
The nondimensionalized equation becomes
6C 162C 6C
=— —— D l-C 6.9
av Peax2 ax+ a( ) ( )

93

In this equation, Pe is the Péchlet number which represents the ratio of the advection rate
to the dispersion rate. Do is the Damkohler number which represents the ratio of the
mass transfer rate to the advection rate.

The maximum and minimum relative error (emax , 8mm) of the implicit transport
code was evaluated from the analytical solution for the soil and fluid characteristics listed
Table 6.1 where:

 

 

Canalyn'cal _ C'implict model

 

.100 (6.10)

g :
I Canalytr‘cal

Figure 6.1 compares the implicit transport numerical code to the analytical
solution for a 15 cm column, with a spatial grid of 0.5 cm, Da of 1, Cr of 0.5 and a 0)
value of 0.5. In this study, the concentration of NAPL in the surfactant enhanced
aqueous phase predicted by the analytical solution and the numerical code were compared
at increments through the column. The results indicate that the numerical code can be
used to predicted surfactant enhanced dissolution of the trapped residual NAPL. The
code was also evaluated to minimize the error between the numerical code and the
analytical solution over a range of Cr, 0) and Da to further evaluate its applicability.

The first of these studies focused on evaluating the model over a range of
velocities. In this study, the velocity was varied while maintaining constant values of Cr
= 0.3, L = 15 cm, Ax = 1 cm, 0) = 0.6 and k, = 0.397 hr'l. Since the velocity is a factor in
both Cr and Da, At was varied to maintain a constant Cr and 0.5 3 Ba 5100). The
results of this study are shown in Figure 6.2. The maximum relative error of 7.48%
occurred at Da = 50, whereas a minimum relative error of 0.004% occurred for Da =
0.01. These results suggests that the model can be used over a range of velocities.

The evaluation was extended to evaluate the error over a range of weighting
values (0.55 co $1) and a range of Cr (0.3 3 Cr 3 0.7). Figure 6.3 summarizes the

maximum relative errors found in these studies. From this figure, the maximum error is

94

shown to be minimized (5.58%) at a or value of 0.6 and a Cr of 0.3. As a result, these
values will be used in further simulation studies using the 1-D implicit transport model.

Two further studies were conducted in order to evaluate the use of the implicit
transport model for a wider range of Da with respect to various mass transfer coeflicients
and Fe with respect to a range of dispersivity values. Table 6.2 lists a range of k1
measured in other studies which assumed that the mass transfer is adequately described
by the same linear driving force model (Borden and Kao 1992; Priddle and MacQuarrie
1994). Using an average interstitial velocity representative of naturally occurring
velocities in sandy aquifers (v = 0.1 cm/min) and a higher velocity representative of low
flow pumping conditions (v = 1 cm/min), the maximum and minimum error between the
numerical code and analytical solution was predicted for this range of k,. The system was
evaluated at a length of 15 cm, spatial discretization of 1 cm, a Cr of 0.3 and a 0) value of
0.6 while assuming a constant value of dispersion. The maximum relative error is below
4.27% for Do < 100 (Fig. 6.4). However, when the values of dispersivity were varied for
naturally occurring aquifer material (Mercer et al. 1982) resulting in 10'2 < Pe < 104,
much larger errors are present (Fig. 6.5) for the same conditions, assuming a constant k,
of 0.387 hr'l. If the acceptable maximum error is 10%, this study suggest that this model
may be limited to Fe 3': 100. A firrther optimization study could be conducted which may
determine that different values of Ax, co and the Cr decrease this error. However, the
objective of this study is to evaluate the dissolution of a single component of gasoline
(dodecane) into a nonionic surfactant (5% Witconol NP-l 50) fi'om a sandy aquifer
material. As such, error analysis of the code will only consider optimizing these values
with respect this study.

Two early versions of the 1-D transport model developed evaluated the use of an
explicit scheme using both the Quick formulation (Leonard 1979) and a centered finite
divided difference approximation for the advection term. In both codes, the centered

finite divided difference approximation was used for the dispersion term. A fixed step

95

size 4th order Runge Kutta method was employed to solve the resulting ordinary
differential equation. Semprini and McCarty (1991) report problems with the Quick
Method near the influent end when high concentration accumulated. The increase driving
force of dissolution at this end for the system modeled resulted in errors which may be
similar. The centered finite divided difference approximation of the advection term

resulted in large oscillations around an average value at later time.

6.4 Simulation Studies Using l-D Implicit Transport Model

The 1-D implicit transport model was developed to predicted the behavior of a
nonionic surfactant (5% Witconol NP-ISO) which enhances the solubilization of trapped
residual NAPL into the aqueous phase. The numerical code was compared to an
analytical solution to evaluate the minimum and maximum relative error during early
pumping, assuming that the distribution of trapped residual NAPL remains relatively
uniform. However, the dissolution of NAPL into the surfactant enhanced aqueous phase
is a rate-limited process which results in an uneven distribution of trapped residual NAPL
in the column over time, limiting the use of the analytical model. Figure 6.6 illustrates the
removal of NAPL over the column length at specified pore volumes (2, 20, 40, 60) using
the implicit transport model for Cr = 0.3, 0) = 0.6, Da = 10, L = 15 cm and Ax = 1 cm.
Clean-up occurs more rapidly at the influent end resulting in an uneven distribution of
trapped residual NAPL remaining in the soil as the remediation progresses. Figures 6.7-
6.8 show similar results for the same conditions but at a higher velocity (Da = 1) and a
lower velocity (Da = 100). Changing the velocity decreases or increases the residence
time of the surfactant. However, the stronger removal at the emuent end is still evident.
Figure 6.9 illustrates the mass transfer over time by solving the ordinary differential
equation in Eqn. 6.2 for no flow conditions.

x— — _
;_J_k,(c, C) (6.11)

96

_C_. = l- e-klt (6.12)

S

In this figure, the mass transfer is stronger at earlier times when the aqueous phase
concentration of NAPL is lower, asymptotically approaching a normalized value of
concentration of approximately 0.8 after about 12 hrs of residence time. Figure 6.10
illustrates how the residence time decreases for decreasing Da values (increasing
velocities), resulting in a range of normalized effluent concentrations that decrease as the
Da value decreases. In the lower values of Da, equilibrium solubility of NAPL is not
achieved by the surfactant enhanced aqueous phase. This inefficiency is illustrated by the
longer clean-up times (e. g., 1* E 88 for Da = 100, t* E 163 for Da = 1).

As the trapped residual NAPL is removed from the soil, the increase in the
effective aqueous phase porosity will result in a decrease in local flow velocity and
subsequent increase in local residence time of the surfactant enhanced aqueous phase.
The implicit transport model was used to evaluate the significance of this local change in
interstitial velocity. In the first two studies, velocity was held constant and predicted

based on either total porosity, n,

n=-—"— (6.13)

or the initial effective porosity, ne,

V..-V..
n:—
V

t

(6.14)

where Vv is the volume in the voids or pore space, V" is the volume of trapped residual
NAPL, and V, is the total volume of the soil and pores spaces. By predicting the velocity
based on total porosity, the model assumes that the trapped residual NAPL does not
influence the permeability of the soil. By predicting the velocity based on effective
porosity, the model accounts for the reduced permeability the trapped residual NAPL In

a third study, velocity was predicted at each time step within each element based on local

97

changes in the effective porosity as the residual NAPL volume is reduced due to
surfactant dissolution. The results presented in Figure 6.11 suggest that the solution
based on local changes in velocity differs significantly from the solution based on a
constant velocity predicted by either total or effective porosity. As such, local changes in
effective porosity will be incorporated into the code to predict local velocity and
residence time.

A numerical simulation was compared against actual data collected from the
effluent end of a column flushed continuously with a 5% solution of the nonionic
surfactant (Fig. 6.12). The model does not predict the trend in the experimental data at
later time which can be explained in part by the limitations in the use of the lumped mass
transfer coefficient. This coefficient is a function of both the specific interfacial area
between the NAPL and the aqueous phase and a measure of the mass transfer across this
interface. Not only is the actual internal contact between the aqueous phase and the
NAPL difficult to quantify experimentally or mathematically due to the complex size and
shapes of the ganglia, but the interfacial surface area changes with time as the NAPL
dissolves into the aqueous phase (Wilson et al. 1989; Miller et al. 1990). Miller et al.
(1990) and Powers et al. (1991) evaluate the mass transfer and the change in interfacial
area as a function of measurable parameters of the soil and fluids. A second limitation of
the model may exist if preferential flow paths or pore constrictions are present within the
soil prior to or during treatment of the contaminated soil with surfactant. The model
assumes a uniform distribution of trapped residual NAPL and a constant interstitial
velocity within an element which does not account for the aforementioned conditions.
However, the data compares more favorably to the implicit transport model than
comparable data in a similar long term dissolution study evaluating an equilibrium model
and a kinetic model developed by Priddle and MacQuarrie (1994). In their study, the

lumped mass transfer term was also measured in a flow interruption study.

98

A higher value of Da (i.e. 100) directly translates into lower pumping rates but an
increase in residence time when compared to a lower value of Da (i.e. 20) which
translates into a higher pumping rate but less residence time (Fig. 6.13). However, both
cases will result in 70-90 pore volumes of contaminated water based on the assumption
that all of the enhanced aqueous phase is recovered and that it is not diluted by ambient
groundwater. The validity of these assumptions will be addressed in the 2-D model study
which evaluates the flow paths of the injected surfactant enhanced aqueous phase in the
next section.

The numerical result of the long term dissolution study of the column experiment
was used to evaluate error propagation in the model. Assuming that the mass transfer
coefficient, residual NAPL saturation and dispersivity are independent, simulations were
modeled changing these values in different combinations to the average value plus the
error associated with the variable. The values used in each simulation were obtained fi'om
. Table 6.1 and shown in Table 6.3. The maximum relative error between the simulation
with average values and the simulation with the values shown in Table 6.3 is recorded in
the fourth column. The value of dispersivity was varied to include dispersivity in Ottawa
sand with and without trapped residual NAPL, resulting in a 0.6% and 2% relative error
respectively. This indicates that error in dispersivity does not have a strong influence on
the model for the system under consideration. The largest error (94%) occurs when the
error in the saturation of trapped residual NAPL is evaluated. It is important to note that
this error occurred at later time, where the predicted values of NAPL concentrations are
relatively small. When the error is propagated for all three terms, the maximum error is

10.3%.

6.5 Verification of 2-D Numerical Code

The potential surfactant enhanced remediation of an unconfined aquifer with a

region of trapped residual LNAPL contamination near the capillary fiinge was firrther

99

investigated using a United States Geological Survey (USGS) model for saturated and
unsaturated flow and t_ra_nsport, SUTRA (Voss 1984). The simulation models a vertical
cross section of an unconfined aquifer across the saturated and unsaturated zones (Fig.
6.14). The lower and upper boundaries are no flow boundaries whereas the side
boundaries are constant head. The region below the water table is approximately 0.7 m
wide, with a capillary fiinge approximately 0.2 m wide. The remaining upper region of
the aquifer is the unsaturated vadose zone. The 0.8 m x 0.3 m contaminated region is
near the center of the aquifer extending through the capillary fiinge and below the water
table. These dimensions were chosen to simulate a physical lab scale model which is
available to conduct supporting experiments. It should be noted that the prototype scale
could easily be increased.

The aquifer is divided into rectangular elements 0.1 m x 0.1 m except in the
region near the capillary fringe. In this area, the aqueous phase saturation and relative
permeability may change from saturated to near residual Saturation values within one or
two element lengths. To account for this, the vertical increments are reduced to 0.02 m
in this region. Element and node numbering are show in Figures 6.15-6.16. The
pressure-saturation and relative permeability relationships in the vadose zone were
initially estimated based on van Genuchten's equations (Chapter V). A pressure-
saturation curve using these values is shown in Figure 6.17.

A steady state flow field is generated for each well scheme. To verify the model,
the location of the top boundary of the saturated zone was compared to an analytical -
solution (Wang and Anderson 1982 ) using the Dupuit assumption of horizontal flow

where:

1(5sz
— =R 6.15
2 5x2 ( )

 

100

In this equation, K is the hydraulic conductivity of the saturated soil, H is the hydraulic
head, x is the length and R is the recharge rate which is zero for the solution being

considered (Fig. 6.18). The general solution is:
H2 =alx+a2 (6.16)

where a 1 and a2 are constants determined by substituting the constant head boundary
conditions at each reservoir. In the analysis, a 0.01 gradient was maintained between the
reservoirs, varying the LHS head between 1-0.2 m in 0.2 m increments. Figure 6.19
shows the results of the comparison between the numerical code and the analytical

solution where the relative error, e, was determined as:

 

 

e: IHWM—HWWIJOO (6.17)

H analytical

The 0.1% error in the specified heads is the result of error in the code which only
maintained accuracy to 5 significant figures. The largest error, approximately 0.27%,
occurred at the smallest elevation of the reservoirs evaluated. These results verify the use
of the flow model in SUTRA.

The transport model was compared against an analytical solution for an
instantaneous point source at the origin of an x-y coordinate system at time t = 0 in a
saturated, homogeneous, isotropic medium (Fischer et al. 1979; Freeze and Cherry 1979).

The initial condition can be expressed as:
C(x,y,0)=M6x6y (6.18)

where M is the mass of the contaminant introduced at the point source and 6x and By are
the incremental thicknesses of the injection point. As the contaminant mass is transported
by the flow through the porous media, the concentration of mass at time t is given by:

2
_ t 2
C: exp _(x v) — y (6.19)
47mb,/D,D, 40,1 419,1

 

 

101

where n is the porosity, b is the thickness of the aquifer and D, and Dy are the
coefficients of dispersion in the x and y directions.

The numerical code simulated the injection of an instantaneous point source by
injecting 0.005 kg of contaminant mass at a center node of a saturated 1 m x 2 m aquifer
in 10 seconds. Figure 6.20 shows the comparison of the analytical solution with the
numerical code with contours based on the fraction of mass 10 minutes after the
instantaneous injection of the solute mass. In the model simulation, the grid spacing was
maintained at a constant value of 0.1 m. The simulation assumed negligible molecular
dispersivity (D* = 0), longitudinal dispersivity of 0.1 m (01,) and transverse dispersivity (a
,) one-tenth the value of the longitudinal dispersivity. These values are within the range
of values of dispersivity reported for sandy homogeneous soils measured in both the
laboratory and the field (Freeze and Cherry 1979). Additional data and parameters used
in SUTRA are summarized in Appendix D.

The numerical transport code predicts the transport of the solute more accurately
fiom the middle to the outer edge of the plume as indicated by the 1% and 3% mass
fraction contour lines. At the injection point, as indicated by the 5% contour line, there is
less agreement. However, the results indicate that the numerical code reasonably predicts
solute transport as modeled by the analytical solution.

Based on the results of the comparison of the flow and transport model used in
SUTRA with the two analytical solutions, SUTRA was selected to further evaluate the

use of surfactant enhanced remediation.

6.6 Two Dimensional Model Simulations

The primary objective of applying hydraulic controls is to deliver surfactant into a
contaminated region where the relative permeability is reduced, to ensure efficient contact
time of the surfactant with the contaminated region, and to recover contaminant laden

surfactant enhanced groundwater. The evaluation of well systems, designed by the

102

placement of wells and pumping rates, will be based on these considerations. The
geometry of the system is illustrated in Figure 6.21. Dimensionless numbers, based on
the dimensions of the contaminated region with respect to the aquifer domain will be used
to indicate the placement of the wells.

Prior to pumping, the flux rate per unit width through the contaminated region is
Qa. Subsequent flux rates per unit width of the source and sink wells (Q,) are normalized
by Qa-

7:9; 620
Q. Q, (- )

The distance to the source and sink wells in the x-direction fi'om the centerline of the

contaminated region, d,, is normalized by the length of the contaminated plume, I.

x; = -——'- (6.21)

The distance to the source and sink wells in the y-direction from the upper boundary of
the contaminated region, 6,, is normalized by the width of the contaminated plume, w.
y;- =3 (6.22)
w
The dimensions of the contaminated region are compared in the aspect ratio w'

where:

w' = 1; (6.23)

A summary of the well schemes evaluated in this research is presented in Table 6.4. In
the case studies presented, the aspect ratio was maintained at a constant value of 0.375.
An initial simulation study assessed the effects of the reduced permeability and
reduced effective porosity resulting from trapped immobile phases at the capillary fringe
on surfactant flow into the contaminated region. Soil and fluid characteristics were

maintained to represent the conditions used in the previous 1-D study and experimental

103

work. In the contaminated region, the permeability was reduced to 0.5K and the effective
porosity to 0.8211. A 6% surfactant solution was injected at a rate of 0.5 kg/s (0.5x10'3
m3/s) and monitored over time as it migrated between source and sink wells of equal
strength (Case la; Table 6.4). The results were compared to a similar simulation
conducted for a completely homogeneous aquifer (Case lb; Table 6.4). Figures 6.22-
6.23 compare the migration of the plume both cases, mapping the surfactant
concentration contour plots from 0.01-0.05 (l%-5%) at t = 1.67 hr., 3.3 hrs. and 5 hrs.)
The concentration contour plots were developed fi'om data generated by SUTRA and
imported into Surfer which incorporated kriging (Golden Software; Golden, Colorado).
The results indicate that the injection is slightly retarded in the study where the relative
permeability and the effective porosity is reduced, but significant by-passing of the
contaminated region does not occur. The results of the two simulations is further
compared in Figure 6.24. In this figure, the 0.05 concentration contour (5%) is
compared at 5 hrs. further supporting the conclusion that no significant by-passing is
occurring.

The contaminated region was subdivided into three subregions, each 0.1 m x 0.8
m, in order to fiirther evaluate the flow characteristics of the surfactant enhanced
dissolution of the trapped residual NAPL. The upper region in the capillary fiinge is
subregion 1, below the approximate location of the watertable is subregion 2, and the
lower region is subregion 3. If flow through the subregions is horizontal, the 1-D
transport model can then be used to predict clean-up in the subregion as a 1-D column.
In such a case, the velocity of the region, incorporated in Da, can be directly used to
estimate the time required for NAPL dissolution (Fig. 6.13). In Table 6.4, Ba in the three
subregions is shown as Da(1), Da(2) and Da(3), respectively. Prior to pumping,
estimates for flow variations in these 3 regions were made for the case where the relative
permeability and effective porosity were reduced as well as for the homogeneous aquifer
(Case 8-9; Table 6.4). The coefficient of variation (the standard deviation divided by the

104

average value) of the velocity within a region was estimated based on the magnitude of
the velocity at the centroid of each element. The coefficient of variation is indicated for
each subregion in Table 6.4 as CV(1), CV(2) and CV(3), respectively. In the
heterogeneous case, the CV in subregions 2-3 is 0.06 compared to 0.003-0.002 in the
homogeneous case, indicating a greater variability in the velocity for the heterogeneous
region. In the region of the capillary fringe (subregion 1), the CV is much higher (0.32),
indicating a significantly higher degree of variability in velocity. Above this region, the
relative permeability of the aqueous phase is significantly reduced. As a result, the top of
the capillary fiinge is similar to an impermeable boundary where streamlines converge.
Velocity vector plots, generated by the supplemental program included with the SUTRA
documentation (Souza 1979), further illustrated these differences in Figure 6.25-6.26. In
these plots, the velocity vectors are scaled by the maximum velocity vector while
indicating the direction and magnitude of the velocity. The smaller grid in the capillary
fiinge results in a higher concentration of velocity vectors, but does not necessarily
represent an increase in flow or velocity. The area of reduced relative permeability and
effective porosity is shown by the shaded region in Figure 6.26, which is not included in
the completely homogeneous aquifer studied in Figure 6.25.

Figures 6.27-6.28 show the velocity vector field for Case la-b, where source and
sink wells of equal strength are used to inject a surfactant. As shown in Table 6.4, Ba is
reduced in the three regions, indicating an increase in the local flow velocity. Variations
in the upper region are slightly increased from 0.32 to 0.34-0.35 (6-9%). The coefiicient
of variation for the velocity in the middle region is significantly increased in the
homogeneous case (1b) from 0.003 to 0.08 and fiom 0.002 to 0.08 in the lower region.
In the heterogeneous evaluation (Case la), the increase in variability in the middle and

lower regions is between 33-50% (change in the coefficient of variation = AC.V. = 0.02-

0.03 respectively).

105

Additional well schemes were evaluated (Table 6.4). Concentration profiles at t =
50 hr. were used to measure the effective zone of capture of the surfactant plume. In
these studies, dispersivity and molecular diffusion were considered negligible using a
value of 0.5 for the weighting parameter UP in SUTRA. This was done in order to
monitor the sharpest concentration gradient, as suggested in the manual (Voss 1984.
p. 10).

The concentration contour graphs were evaluated by comparing the area captured
by the 5% surfactant plume (A 5%), the total area of the contaminated region (A c), and the
area of the contaminated region not captured within the 5% surfactant plume (Au). Two

measures of the zone of capture were developed based on the ratios of these areas.

 

Aso/
: -——° 6.24
51 Ac ( )
A
6‘ = 1- u 6.25
2 A ( )

0

Values of 81 >1 suggests that the surfactant plume is migrating through uncontaminated
regions of the aquifer in addition to the contaminated area targeted for treatment, where
as 81 < 1 suggests that the surfactant plume cannot treat the entire contaminated region.
This measure of the zone of capture does not evaluate the extent of the contaminated
region left untreated by an injected 5% solution of surfactant, which is provided by the
second measure, 82. This measure of the zone of capture provides an estimateof the area
left untreated or else treated by a surfactant enhanced aqueous phase diluted by ambient
groundwater such that the surfactant concentration is less than 5%, where dissolution of
trapped regional NAPL may still occur. Figures 4.6-4.7 indicates a CMC < 1%,
supporting that surfactant dissolution is not limited to regions of 5% concentration.
Analysis based on the dissolution of NAPL and the rate limited mass transfer variations

based on surfactant concentrations would require a more complex model and additional

106

experimental work beyond the objectives of this research. Despite these limits, this
simplified analysis provides a valuable estimate of the evaluation of hydraulic control.

In Case 2, screened injection and recovery wells of equal strength
(Q; = Q; = 1.49) were placed below the water table and outside the predicted
contaminated region. The resulting concentration contour plot and velocity vector plot
are given in Figures 6.29-6.30. Based on the concentration contour plot, 81 = 1.73, as
evident by the large 5% surfactant plume which extends below the contaminated region.
The pumping scheme is highly efficient at capturing the contaminated region as evident by
82 = 0.99. For the pumping strategies studies, this scheme is the most efficient in
capturing the contaminated region within the 5% plume. As state earlier, lower
surfactant concentrations also enhanced the dissolution of the trapped residual NAPL in
the contaminated region. The entire contaminated region in this study is captured by the
surfactant plume < 1%. However, the recovery well may not be recovering all of the
contaminated laden surfactant, and is definitely not recovering all of the surfactant. The
velocity is relatively horizontal in the contaminated region as indicated by the velocity
vector plot and variations of the magnitude of the velocity in the three subregions.
Consistent with the previous considerations of velocity differences in these subregions,
the upper region of the capillary fringe shows the highest variability, as indicated by the
coefficient of variation (CV) in region 1 equal to 0.35, as indicated by CV(l) = 0.35. The
CV in the other two regions is increased compared to Case 8 with no pumping. Because
of the higher flow velocity due to pumping, Da is increased in all three subregions,
compared to Case 8. The average velocity in each of the subregions of this study was
used as input in the 1-D implicit transport code to predict the fraction of trapped residual
NAPL remaining after flushing with 60 pore volumes of surfactant. The results (Fig.
6.31) indicate that the variation in velocity does not result in dramatic differences in

clean-up time.

107

In Case 3 (Figures 6.32-6.33), the strength of the screened injection well was
reduced 50% fi'om the previous study. As a result, the total area of the surfactant plume
was reduced, as evident in el = 0.87, but less of the contaminated region is captured in
the 5% surfactant plume as indicated by 82 = 0.95. The CV in the upper region
increased, but slightly decreased in the lower two regions. The Da increased due to the
reduced strength of the injection well compared to Case 2. The velocity vector plot
indicates that the flow is relatively horizontal, indicating that the 1-D transport code may
be used to predict NAPL removal. The graph of the surfactant concentration profiles
further indicates that all of the contaminant laden surfactant and all of the injected
surfactant is removed from the groundwater environment by the recovery well.

In CaSe 4 (Figures 6.34-6.35), the strength of the screened injection and recovery
wells were equal (Q; = Q; = 1.49) but placed within the contaminated region. In this
study, 81 = 0.91. However the inefficiency of the scheme is evident in so far as 82
decreased dramatically to 0.83. Of the cases studies, this pumping scheme captured less
of the contaminated region than any other. The recovery well does not recover surfactant
which may have flowed through the contaminated region, thereby spreading NAPL
contamination into previously uncontaminated regions of the aquifer. In addition, the
velocity vector plot indicates that the velocity field in the contaminated region is
unidirectional and non-horizontal, negating the use of the 1-D model as a tool to firrther
evaluate this pumping scheme.

In Case 5 (Figures 636637), a smaller screen well is evaluated just below the
watertable but outside the contaminated plume (Q; = Q; =1.49) in an attempt to deliver
less surfactant below the contaminated region. As a result, more of the contaminated
region was captured (82 = 0.97) but the size surfactant plume increased (81 = 1.47). The
velocity vector plot indicates that the velocity is relatively horizontal in the contaminated
region, but the concentration contours indicate that the surfactant is not completely

captured by the recovery well.

108

The lowest injection and recovery well strength was evaluated in Case 6 where
Q; = Q; = 0.3 (Figs. 6.38-6.39). At this low flux rate, the surfactant was slowly injected
into the groundwater system. As a result, the 5% surfactant plume is diluted by ambient
groundwater. The velocity vector plot suggests that a small degree of by-passing may
occur, as indicated by the velocity vectors outside the lower left side of the contaminated
region. The zone of capture is extremely small compared to the other situations studied
and was not evaluated in this case.

Case 7 (Figures 6.40-6.41) evaluated a pumping scheme where the surfactant was
injected between injection and recovery wells of equal strength outside of the
contaminated region at a flux which maintained flow through the contaminated region
slightly above the flow that would be in the region without pumping (Q; = Q; = 1.19). As
illustrated in Figure 6.40, the 5% surfactant plume captures most of the contaminated
region, but extends over a larger region. In addition, the injected surfactant is not
captured by the recovery wells, as suggested by the concentration contour plot and the
velocity vector plot .

In all of the cases studied, additional evaluation based on the coefficient of
variation of the velocity in the subregions is summarized in Table 6.4. This table also
summarizes the results of the zone of capture measurements. Comparisons of this table
with the concentration contour plots and velocity vector plots indicate that the most
efficient pumping scheme is Case 3, where screened injection and recovery wells are

placed outside the contaminated region, but pumped at unequal strengths such that
Q,’ = 0.5Q, = 0.75.

6 .7 Summary
A 1-D implicit transport model was developed and used to evaluate the use of a
surfactant enhanced aqueous phase to promote the dissolution of trapped residual NAPL

in conjunction with 2-D modeling using the USGS saturated-unsaturated transport model

109

SUTRA. The primary focus of the research was to evaluate the use of hydraulic controls
to emciently use surfactants to enhanced the dissolution of trapped residual LNAPL near
the capillary fiinge. The study was conducted in a homogeneous unconfined sandy
aquifer with a region of trapped residual LNAPL contamination near the capillary fringe.
The vertical cross section of the aquifer extended 1m x 2m with a 0.8 m x 0.3 m
contaminated region. These dimensions were selected based on an existing physical
laboratory scale model, but the prototype scale could easily be extended.

Several different well schemes based on well location and pumping rates were
compared- Surfactant concentration profiles and velocity vector field provided the basis
for selecting a pumping strategy based on 1) delivering the surfactant to a contaminated
region; 2) ensuring sufficient residence time between the surfactant enhanced aqueous
phase and the contaminant to minimize the cost of the surfactant and the amount of
wastewater to be treated; and 3) recovering the contaminant laden surfactant enhanced
groundwater.

For the small aquifer studied, a surfactant enhanced aqueous phase flowing at
local flow velocities will not reach solubility concentration of LNAPL after migrating
through the contaminated aquifer. If the local ‘flow velocities are increased due to
pumping, the dissolution of LNAPL into the surfactant enhanced aqueous phase may be
further decreased. As a result, the primary use of the hydraulic controls were to promote
the delivery of the surfactant into the contaminated region and the recovery of
contaminant laden surfactant enhanced groundwater. The strength of the wells and the
location of the well with respect to the dimensions and location of the contaminated
region are summarized in nondimensional numbers in Table 6.4. The results indicate that
the most eflicient design maintained a constant injection rate below the rate of the
recovery well set outside of the contaminated region. The results support the need for
further research in the use of surfactant enhanced groundwater remediation of trapped

residual NAPL, including addressing different aquifer soils or dimensions.

110

In less permeable soil, slower velocities may result in a more optimum use of the
residence time through the contaminated region. The surfactant may also be more
efficient when evaluating a larger contaminated region. In both cases, the maximum
solubility may be achieved prior to recovering the contaminant laden surfactant enhanced
ground water. In these situations, the flow may need to be increased to optimize the cost.
Further evaluation of the capture zone would also need to be conducted.

In this study, heterogeneity was limited to the inclusion of a contaminated region
where the relative permeability of the soil is decreased to 0.3 to 0.7 times Ksat due to
trapped residual NAPL. In the field, heterogeneities are also present due to silt and clay
strata that may reduce the permeability by a factor of 1000 to 10 million (Mackay and
Cherry 1989). If the site investigation to determine the aquifer soil characteristics was
limited due to economic constraints or limitation from existing structures (e. g., buildings),
such strata may not be factored into the evaluation of the hydraulic controls. Based on
these considerations, further evaluation should also incorporate changes in the efficiency

of the hydraulic controls which may occur due to undetected lenses of clay or silt .

111

 

‘19 " —0— ANALYTICAL sour.
03 -. <> mmcrr sour.

 

 

 

0 0.1 0.2 0.3 0.4 0. 5 0.6 0.7 0.8 0.9 l
X/L

 

 

 

Figure 6.1. Implicit numerical solution compared to analytical solution ( Cr =0.5; L
=15 cm; a) = 0.5; Ba =1; Ax = 0.5 cm)

 

 

 

 

 

 

 

"r
o
1.
0 o
o
5‘ o
o
A
35"
E o
“' o
o
is 0 'Maximum
°Minimum
2.!
1
0
0L t 1‘ H , .
0 10 20 30 40 50 60 70 80 90 100
DI

 

 

Figure 6.2. Maximum and minimum relative error over a range of velocities as
indicated by the range of Da (Cr = 0.3; L = 15 cm; a) = 0.6; Ax = 1 cm).

112

 

 

 

 

 

 

100
10
A
a"
v
a —v—Cr-o.3
a +Cr-05
“’ 1 +Cr-0.7
01 I *7 1 I I ‘ I f T fi
04 05 06 0.7 08 0.9 10
(D

 

 

 

Figure 6.3. Summary of study to optimize the Courant number over a range of to
values (L = 15 cm; Ax = 1 cm).

 

 

 

 

 

 

 

100 a
: A
A 9 MAX ERROR (v=l cm/min)
o\'
E 10 0 MIN ERROR (v=l crn/min)
E A MAX ERROR (v=0.1 cm/min) . 8
E 1 A mm ERROR (v=0.1 cm/min) a
j a
a O. 1
> o
0.01 i i i i i
0.01 0.1 1 10 100 1000
DAMKOHLER NUMBER

 

 

 

Figure 6.4. Relative error over a range of Da numbers represented in previous
table (Cr = 0.3; a) = 0.6; Q = 0.5 mein; L = 15 cm; Ax = 1 cm).

113

 

 

 

 

 

 

1000 OMAXERROR(v-O.lemlrnin)
o MINERROR (F01 mm)
3 100 a a a AMAXERROR(v-lanlmin)
3.4 AMINERRon-ran/tnni)
‘ O
O
a 10 A a
H . 8
E 1
5 O
a o o o o o
0.1
‘ A A A A
001 t t i a i i i i
0.01 0.1 l 10 100 1000 10000 100000
Pe

 

 

 

Figure 6. 5. Relative error over a range of Pa numbers (Cr = 0.3; a) = 0.6;
Q = 0.5 mein; L = 15 cm; Ax = 1 cm).

 

 

 

 

 

 

 

 

 

1 ‘ .—---———'—“'—-—__.._...:-T-T7-T="—" ............ h" ...............
0,9 dr- ,// “....M'Wl ----- 1,.r"" .1", ------
0.8 -U- ’3'.” l/,’ 31""
0.7 a. [I
l.‘ / 'I
0.6 ~- - — — — t‘ = 2
s 0.5 . 1' ," --------------- t. = z)
3 0'4 1.. '1' // ------- t. = w
0-3 " 2’ - ---------- t“ = 60
0'2 ._ ",1
0.1 ~-
0 ~ *fi t 4' i 41- i ----- + ----- - i i a ---i
0 0 1 0 2 0 3 0 4 0.5 0 6 0 7 0 8 0 9 1
X/L

 

 

 

Figure 6.6. Fraction of trapped residual NAPL remaining in pores spaces over the
length of the column over time (Da = 10; Cr = 0.3; a) = 0.6; L = 15 cm; Ax = 1 cm).

114

 

 

 

 

 

 

 

a 1‘-——-—-—~*"*————————“——~——‘—‘f

- 0.8 - """""""""""""" g - _____ - - ~, ......

30‘“ ”..-!” ...... I” _____ g ----------------

.4 0.6" ” ,,,,, , ..... I” --------- t‘-2

Q 0.5 .. f, ..... , "’ .................... ,.=,,

z 0.4 + "

‘8 0.3 .. -------- t‘=40

.s ,2 .

g . _ .............. 9:60

i. 0.1 --

k‘ 0 t t t . %
o 0.2 0.4 0.6 0.8 1

X/L

 

 

 

Figure 6.7. Dimensionless analysis of surfactant enhanced dissolution of trapped
residual dodecane (Da = 1; Cr = 0.3; 0) = 0.6; L = 15 cm; Ax = 1 cm).

 

 

.................... tfi=20

- --------- t"=40

L
v

 

 

[ :3: I" I - """""""""" t. = 60

 

FractionofNAPLRemaining
PPPPPPPPP
Ou—Nwhuc\~IOOVOI-'

 

0.2 0.4 0.6 0.8 1
X/L

O

 

 

 

Figure 6.8. Dimensionless analysis of surfactant enhanced dissolution of trapped
residual dodecane (Da = 100; Cr = 0.3; co = 0.6; L = 15 cm; Ax = 1 cm).

115

 

1.0-

0.8 -

 

0.6 -4

C/c,

0.4-J

0.2-r

0.0 .

 

1 I T I V l 1 I V T T I

8 1o 12 14 16 18
time (hrs)

@-
Oj-I

 

 

 

Figure 6.9. Rate-limited dissolution of NAPL into surfactant enhanced aqueous
phase.

 

 

 

 

 

 

 

1 -r
- -------------- Da=100
—————— Da=10
08 ‘ :7::T:::":“':‘:':”\‘ """"
§ ‘ - . ‘\. - --------- Da=4
Q A A A A A \,\ \ \
“ A ‘\
A ‘\ A Da=2

g 06 i an;
E o o O o <> <> <> 0 0‘60 o O Da=1
H E\\\.‘!A O
30» ‘=.\‘x. A O o

i 0
E g 1 ‘.‘ A 0
e '= l, x
z . E x A O

0.2 r l, \‘ <>
l “ A
'=. ‘. 0
0 t + f t f 1 t #13 t w .
0 20 40 . 60 80 100 120 140 160 180
ti

 

 

Figure 6.10. Normalized effluent concentrations over a range of Da.

 

116

 

 

 

 

 

 

 

 

1 “ OTOTALPOROSITY
[-1 0.9
Z A EFFECTIVE POROSITY
a 0.8
p g 0 7 O VARIABLE POROSITY
E = ‘
H E 0.
G
g E 3.4 0 ’ A A A
. O
z 0.6 A A
8 0.3 i' O . A A
«)- 0 .
g 0.2 o . A A
0.1 ”J“ O O A
0 . fee a .
0 50 100 150 200

 

 

 

Figure 6.11. Comparing effluent concentrations when velocity is determined from
total porosity, effective porosity and variable effective porosity.

 

p—s
J.

 

.°
xo

X DATA

.9
co

------------------ IMPLICU‘ CODE

 

 

C
Q
X
X
X
X

 

°.°.°.°..
~63th
X

NORMALIZED EFFLUENT
CONCENTRATION
. o o
o at 05
X x
X x
X
x
Xx
x
x
X
X

 

 

50 100 150
t t

O

 

 

 

Figure 6.12. Experimental and implicit modeled effluent concentrations of
dodecane in a 5% nonionic surfactant enhanced aqueous phase.

117

 

 

m-r

 

 

 

 

 

Figure 6.13. Pores volumes needed to reach MCL over a range of Damkiihler
numbers.

118

£39.. aoumsua $.82 .36 9.53

 

 

x%
the

AWGNDOH (IVEIH .INVLSNOC)

- W

i so

>M<QZDOm mqmgmgu
a N

o

 

 

 

mnmfimfig

5.0a

  

th<m2<w {coo .A

 

m 0255 H>M<AAE<U

mZON mmOQ<>

 

 

 

 

mo<gm QZDOMO
Am

aw

InI

AWGNHOH CIVEIH lNVlSNOf)

am
on]

kw

 

 

119

a: .o x E— .o
mFZmEmt—m «mp—.0 ~ 2

Euod x a: .c Emqum
magnum OH HOZ ZOHOME ngw

.25 EecaeaneEeoazaam 2.. ... 35.5% .2.» as»;

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

mam nmn non an" .en man an" n_u no“ _m_ n6_ n4_ eu_ no_ "a no no an a_ A
Van eun ”on can an" an“ on" n_n eon ~w_ we. ea_ «a. o__ «a as on an on a
man nun don _a~ an" mam emu a_~ _ou mm_ ne_ nv_ n~_ ___ me no an an _« n
eon ”an o_n «an an" on" an“ em" «cu vn_ e62 «v_ on_ u__ «a an an 64 «a a
hen dam _~m new no” Emu an" _~N men nm_ ne_ ee_ _n~ n_~ no on an av an n
«on can u_n van on" an“ can «on we" ew_ me_ en_ and v__ ea an 6e an an e
aam _mm m_m no" on" an” _v~ an” «on EMA oe_ ~n_ mm. n__ on an .6 no «N n
.enm. ~nn., can. .eou. _nn~. ,oen. «an. .4un., ecu m.em_ _.onfi .,un~ ,.¢na ..e~ A».aau weom.n .ne_.. 4 A. .ona awe.m
. Ann . man .nzm no" aeu.v zen. no". an" .56" an. An~.. mn~;..nm~_..e~—. .na _ .awmt.vme,..tuv., «an, o.«.
«an «an eam man can «on aeN, on"- we". ea“ . aha. wax. on" .wxa . 00.. . we..._ae. .evw-. an. _ oag
,nun nmm ,e_m am". _m~. men no”. EN" .aeu .._a_ .mEA.  aa2M an“ e__ ”ox... no. ..ne. .ev ; «a..._aa.
enm 6mm nan. eon an". eon ea" nun ox" ..nez and on. 4 nm__ on_ ”No“ .4» . ee ,. we. on...s-.
«an Ema dam .on .mwu he” he" an" 22" max ne_. on“ am_ ~N.H mo_, .nm. . he . .ae. .ana...n2_
enm man, can . «on .amn .6en «on, cam ".m. aa_ on“. «an. ee~ «NA V,ao—, .ea . we. ..on ;. «n.. .«~.
Ann «an "an men “an no” on" __m~ ..n_~ ne_ and and Ae. -,nn_. no_. a» .ae ~n,. .nn H..n~,
.wnn can nan. eon” .emn we" on" ~m~,. «AN. ea_ «e. co.. Ne. .v~_, .eo~...am, ..oe.-..~n.a ,4n.h te_x_
.anm_ ._an_ nan. .mon. than . no" _.~nn.. nmfi., n_ua.-na_. neg, .~e~.. noa.a «NA» ahead .au _..Ln _ .nnaw. «no . nan
can «an van eOm ”an on” an" an" e_n no. on_ «e. ‘42 en. mo_ ea an an em a.

 

 

120

E: .c x .56
mEmEmAm mmZFO 13¢.

Enod x a: .o 3.75—2me
m4<UmOH
HOZ ZO—Omd Omazm

3n

awn

mam

van

3m

3m

hum

man
an
can
an
Nam
mam
van
non
can
ran
«an

an

.etU Engaganagoaflauam 05 E 8.52 .36 9..—Mr.—

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

Non mvm van mom can hon av" mun ¢_u ~o— Nb— nn— vm— n__ ma hh an an an —
new vVM man can ban awn awn on" -N No_ wh— Vn_ nn— o_— ha «5 an O? ~N. N
vow nvn can hon wwu awn Own gnu N—N ma_ vs— nn— wn~ h—— mm «5 co ~v an n
«ma wvn hum wen amu chm ~nu «an n—n va_ nb_ wn_ hm— m—— ma ow _w «v MN v
own hvm mum mom can ~hu an" and v—u no_ ch— bn~ wn— a—_ co— ~w no mv v" n
hon men awn o—m ~an Nhn nan emu n—N wa~ hb— wn— an— on_ «o— am no vv nu w
wwn own 0mm ~_m mam mun Van nnn e~u ho_ an" any cc~ ~N— NO— ma «o nv on 5
men Onn “mm Nun nmu chm nnn wma h_N ..wm~ ab— cog fiv— NN— no“ cw . no wt .hN w
Ohm nnn «mm m_m van nbm emu hmu w- g.am~ emu .,«w_ «VA .mu— *0“. n» . we , he «N, a
~bm awn mmn v~m non mun hnu. «mm o—N OON ~w~ «om mv_ vn— mom .w« A no..._w¢., rmfl . an
«hm mum vmn .n~m can hhn an" 0mm emu ~9n Nw_ mo~ ¢¢~ and mom hm no ., «V on ~—
mnm vnm nmn w~m ham mhm awn ova "an «on mw~ vw— we“ oN~ ho. an me, an ,nm «—
vrm nun omn h_m wow aha own ~vn Nun non vw~ nw— ,wv_ hnfi wo~ an .ch . an .Nn.. ,n~
urn wnm hmm w_n mow ow” ~w~ Nvu mun. voN mm“ ow~ hv~ mm— mo~ ,ca ,~b ,;nn cmm .v—
ohm ham «an m_m com ~w~ non nvu Van no” ow~ ho~ mv— an~ c- mm Nb an .vn n—
hhm wan omm can gem «mm men «ea nun . ecu b»— we" av— cm_ -~ no nh Vn mm on
”hm .ann own «an «on nun won men can , hen ,:mm~ . mom .on— ,~n~ _N- -.no Vb an on .. b~
ohm own _vn «an non van new wvu baa won «nu or— .m— an— n__ v0 uh en hm «—
omn _en «Va nun vow nun own hVN was mom emu —h_ «n_ mn— v—— ma or hm an a—

 

121

 

0.7 -
0.6 L

0.5 ’

hc (111)

0.4 ’

 

0.3 ’
0.2 "
0.1 L

A 1 A 1 A 1 L J a
000.0 0.2 0.4 0.6 0.8 1.0

Saturation of the Aqueous Phase

 

 

 

 

 

 

Figure 6.17. Pressure-saturation relationship based on van Genuchten parameters.

 

 

 

 

 

 

 

 

 

 

 

Figure 6.18. Schematic illustration of the water table and boundary conditions for
analytical solution of seepage through a dam (based on Wang and Anderson 1979).

122

 

 

 

 

 

 

 

-------- h,=1.o
0.30" ‘ ' ' ‘ h1=0.8
’ . . ---~--~h,=o.6
0'25: ''''' h,=0.4
T h =02
of, 0.20" 1
E ,
m 0.15’
0
..>. i
E 0.10” B
o
m > I. I f“ .‘ . .'
0.05 * 5. j .7
o.oo* ~
00 05 1D 15 20
x-direction (m)

 

 

 

Figure 6.19. Results of the comparison of SUTRA with analytical solution for
determining the surface of the watertable through the aquifer.

 

 

l.“

0.8 1-

O.“ '-

0.“ -

Y-DIRBCI'ION (In)

 

 

 

 

 

 

 

 

0.20 I- .......
0.“ l L l l I l I l l
0.“ 0.20 O.“ a.” 0.” 1.” 1.” 1.“ 1.“ 1.” 2.”

X-DTRFCTION (m)

 

 

 

Figure 6.20. Comparison of numerical and analytical solution for transport of an
instantanteous point source.

123

 

 

C I i c

 

 

 

 

 

 

source 5 H ............. .1. ................................................. u sink
well ‘ well
N d 1 ... d 2

 

 

 

Figure 6.21. Geometry used in evaluation of well placement.

124

 

 

 

 

 

 

 

 

0.20 '-
T- 1.671!

 

 

1"0'0'1 1 1 1 1 1 1 1

0..
0.. O.” 0.“ O.“ 0.“ I.” 1.2. 1.4 hfl 1.“ 2.“

 

 

 

 

 

 

.ea"
_ r-xrw
,. ii-MI. . . . . . '. 1
‘0. O.“ O.“ O.“ O.“ '0“ 1.” 1.“ 1.“ 1.“ 2.“

 

 

 

 

 

 

 

 

 

 

T-51br

 

 

..g l i-o.OI_L L l l l I l I
a” 0.” ... 0.“ .0“ '0“ 1.20 1|“ 1.“ 1.“ 2.00

 

Figure 6.22 Concentration profiles as plume migrates through a region of
contamination where the relative permeability and effective porosity are reduced
due to trapped residual NAPL (Case 1a) at t = 1.67 hr., 3.4 hr. and 5 hr.

125

 

SOURCE NNK

............................................................................

 

T-L7M

 

 

ll'0.01 l I l l l L l l

O.” .
O.“ 0.8 0.“ 0.60 0.9 1.“ 1.3 1.40 1.00 l.“ 2.“

 

 

 

0.8 r
T-13M

 

 

O.” 1 i=0“ 1 1 L 1 4 1 1 1
0.. 0.20 0.“ 0.“ 0.“ ‘0“ ‘0” 1.“ 1.“ ’0“ 2.00

 

 

" 'edu3oe-a-u---.a--\vu-nn ....ln. .....................

 

0.8 -
T-Shr

 

 

1-0.01
.. l I
‘0. 0o” .0“ 0.“ 0.“ 1.“ 1.20 ‘0“ 1.“ ‘0“ 2.“

1 1 1 1 1 1 1

 

Figure 6.23. Concentration profiles as plume migrates through a homogeneous
medium (Case 1b) at t = 1.67 hr., 3.4 hr. and 5 hr.

126

A: 8.5 E... a“ 8:0 .3.— Eseg m «a 2:9:— =e3ah=3=8 new so b.2555 .vue 9..—arm

 

8.0 8; 0o.— 0v.— ON; 8.? 00.0 00.0 0V0 0N0 00.0

8.6

_ n _ . a _ _ _ 8612

Emuh . > .
. w .1 8.6
3 one 208mm €255.28 H
. M /. 19.6
3. one 2282 BomZmeonm .

, , O .1 00.0
....................................................................................................................................................... .. . . M.
............................................................................................................................................ 1‘4... 1..........................I........-t......t..ilth_ 8 .0

225

 

 

 

 

127

Ave 225 a 6:0 no. a... .862 £8; one 2:3...

c5 zofiommak
2 A: 3 I S S 2 ed w to 3 ed
A. A .H A A A A. .. 0 od

 

 

 

Al Al Al al Al Al Al al Al Al ..Art Al Al Al Al A1 Al Al A1 Al
Al. Al Al Al Al Al Al Al Al Al hA.I Al al Al. Al. Al. All Al. A1. All
Al: Al Al Al Al Al Al Al Al A.I Al Al al A1 A1 A1 A1 A11 a1. A1 ..A
All Al 41 41 a1 A1 Al Al Al Al ”A1 A11 A1 A1 A1. A1. A1 A1... “A1... A1 1%
All 61 A, 41 41 Al Al Al Al Al 1 a1 Al Al- 51 A11 A1... A11 “A11 A11 m
..... .5 . .. 1 e o m
h . .. _ A .. A H . m )
... -.. 11-.....” ...... T ..r T “T T “T .l “T T ”1 .II ..II All H ..
RV m A4, 4.! , a1 41 .Al Al .Al Al “.1 A1 “A1 A1 Ha... .1 “Mill M“ A n“ m
. Al a! A 41 Al Al Al “Al Al ”Al A... .1 Al .61. a... a e\ e\ . >\ I..\
f II of . 0| 0| 5| 0| 0| 6| . 0| 0| H 01- Ol. H 61.. .1 H .1 b\ H \
......u......a.........I.......1....».0.. ...... .....1........t......4......n...... ........ .....‘1, ....... . 1..-..............n...-.................r..........a:.‘1..........................1.a.....3................IJ"......H...1.3.33...........O!............a.....l....... ..... . ......t..o......‘............ .....hb ”6°
. a o o a o o o . e I I a I O . 0 v . I I C I

 

 

 

A:

A...» use a 3.6 .e s... .392 3.8.0.» .3.» 9...»...—

Aev zoiomfiak .
2 o; wd wd to No ad

 

128

 

d

 

 

 

 

 

 

. . _ _ _ d d. ._ co
Al Al Al ”Al Al. Al Al- XI AI HAl Al HAl. Au|
5| Al Al Al. Al Al Al Al .Al Al HAI All Al Al.
led .0
fl Al Al Al» AI A... A1\ All Al Al Al All Al Al 4! AI A.| A1l A\u Al. N
_. _ O
T All A.‘ Al Al A\ 01 .1: 4| 0... .1 4| 0.1 t... 4' AI A|.. A\ A‘ A|I l
_ ..od I
_. O
4/ Al Al Al Al 08 ...: 0! 0| «I AI. AI. t... A11 *I. A‘ A\ F\ x N
. _ \/
fl .7 r .1 r H ..H H H n ..H n M ..H ..n ..H ..n .\ m
f A, II A, ll C! b... t3 08 53 £1 9... A‘ 5‘ A‘ L‘ b\ F\ (
. x , r r r - - .. - - .. , x x x x x m a.
. J .. .. .. . . . L

129

A...» 03.5 fl 8.6 .e 8... .82: E83» .2.» 2:5

 

 

 

ON ”A 04 *4 NA CA wd 0.0 Yo Nd Cd
_ _ _ . _ _ ~ _. ._ co
0 o A. A. W A. A. A: A: A: A: A: A. A. A. A. n A. A. A. A.
A. o A A. A. A. Al Al Al N A: A: A: A. f A. A. A. 4 A.
n 1 No
0 Q A. O. 0‘ L\ Al At Al H Al Al A! A! 4: A. I. a. A. O
A A A A. R A\ A\ A. Al Al Al A: A: 4: f i o o o
.. v.0
Q Q | \ ‘ ’ l O O O
O O I 2 $ t O O.
. , o 1 ed
A. w A . s k . . 9
fl A r . a
1D I A K A A I l
u u . m m u u u u u
. . . K A A A A . ,
, , ,, . . . . . we

 

 

 

 

 

 

o.—

130

 

 

 

 

 

" "§&¢

‘ :

—>—>_

—)->§

 

 

1.0

0.6 _

0.2 -

0.0

2.0

1.8

1.6

1.4

1.2

1.0

0.8

0.6

0.4

0.2

0.0

Figure 6.28. Velocity vector plot for Case lb (Table 6.4).

131

A...» 0335 N 8.5 .8.— »o... 38:3 gash—.350

5".» 95w...

 

 

 

 

 

 

®m._. 00; 0?; ON; 00.? om.o 00.0 OV.® 0N.® 00.0 .
_ _ _ _ _ _. _ H . _ ooo
, Sol.
28".» Al>
l ON.®
l 0V.®
.1 00.0
W
I 00.0

 

 

 

132

 

 

 

.992)?

 

 

.
.
u
I I
11“... 0
.
.

 

9

 

 

1.0

0.4

0.2

0.0

2.0

1.8

1.6

1.4

1.2

1.0

0.8

0.6

0.4

0.2

0.0

Figure 6.30. Velocity vector plot for Case 2 (Table 6.4)

133

 

 

 

 
 
  

 

 

 

 

 

1.0”
. Region 1 .

,5 Reglon 2
’g 0'8 """"" Region 3
g b
a:
—l 0.6“
a...
<2 .
E
O 0.4“
r:
0 .
é
a. 0.2-

o 1 l J

(2)0 O 2 0.4 . 0 6 0 8 1 0
X/L

 

 

 

Figure 6.31. Fraction of mass removed in each region after surfactant flushing 60
pore volumes.

134

®®.N

Av.» 03': m 8.5 .5.— 3E ...—3:3 notabaougu

.3.» 2...»...—

 

 

 

 

 

Om; 00.w 0V; ON; 00... 0V.® ON.® 00.0

_ _ _ L _ . . _ 00.0
,Sonm.
affine An|>

.l GN.®

l OV.®

I 00.9

KIV
.... 00.0

 

 

 

135

A...» 03.5 ... 8.6 .8 3... 58> £8...» .9...» ea»...—

 

 

 

 

cN wfi m: v— N_ c; we we to N6 o.o
_ . _ _ .u a _ fl _ 9.0
o A. A. A. A. A. A. Al Al Al Al Al Al Al A. H A. A. . A. A.
A. O. o. A. A. Al Al Al Al Al Al Al Al Al A. AI A. AI A»
. _ ............. ... N.o
O A. A. b. A. A\ A. A. Al Al Al Al Al Al Al Al A. A. A.
9 A A. . R N 4. A. A. A. A.
o . . e R g A A. A. A.
, o .o - ».c
A. .v T t A. 5. Al
14% $ M; 5 Q ‘
. n
. m w m a - n .m
d . . u n .. u w .
. . we

 

 

 

 

 

A:

136

®®.N

.AV.

 

 

 

 

 

 

o 2%: v 030 5m SE 58:8 8255080 .3.» oswfi

00:. 00;. 0?. ON; 00. 0m.® 00.0 OV.O ON.® 00.0 .
_ _ _ _ _ _ _ L . q 000
. . Sol .
Eomne >
I ON.®
,l OV.®
,l 00-0
W
l 0m.®

 

 

 

.c.» 2......» e 3.6 .8 8... .3»: £83» .mn.» 25.:

 

 

 

 

ON ”A o ~ v — N _ 9— w o w o to Nd od
_ _ — q _ _ n _ _ CO
e. f A. A1 Al Al Al Al Al Al Al Al Al Al 4. A. A. a. A.
O. 0. a. A. A1 Al Al Al Al Al Al Al Al AI 4. A. l. t. t.
- Nd
A. A. A. A. A. A. A\ Al Al Al A! A. A. A. A. A. A. A. A.
A. A. A. A. A1 A1 A\ Al Al Al A! A! A: 4. A. A. A. A. A.
"M , l v.0
l
A. A. O. 0. D. 1 f 9 A. A. O.
A. A. O. | l n i A. O. O. l.
.. v.0
‘I ' ’ I O O ‘ .A *
MM A. o A . . . A A R
.. n u u U u u H u o .M
a r A A A . v A A n .
...... : . . ... ..1... .. ”o

 

 

 

 

o.—

138

Av.» asst m 02.0 ..8 .2.— .=_S=3 5322850 .3.» PEEK

0N6

 

 

 

 

 

om; QO. 0?. GN; 00.? 00.0 00.0 0V.® 00.0

_ _ _ _ _ _ A . _ 00.0
wool
Homna +|>

I ®N.®

I OV.®

I 00.0

V
8.0

 

 

 

139

Ava 235 m 3.5 .3.— 3... .580.» £85»

.5.» 2:5

 

 

 

QN «A w; VA NA CA ad v.0 to N6 od
_ _ _ _ a A ~ u 0.6
1 N6
I to
. . . . A K n . .
1 0.0
O ' I I ‘ I .
.m . . . m , V .
n. n H N. M. W n .u
, J . , m6

 

 

 

 

 

 

 

A:

140

®®.N

.93 3.3.5 e 85 ..8 «a... 53:8 53955950 .wmé charm

 

 

 

E 225202
®m.— ®0.F ®V._ ®N.F ®®.F ®mu® ®O.® ®V.® ®N.® ®®.® .
A A _ _ _ q _ . _ 09 ®
. W W :21
.2 0m NH A >
_ _ - 8.9
‘1 0V.®

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

®0.®

®m.®

 

®®.F

 

(w) Momma-A

141

Av.» 0535 e 030 he .2.— ..83> .0323; in.» 9..-Eh

 

 

 

 

 

 

QN w.— A: v; A: ad ed v.0 Nd ed
,. .. . .— A A A. A o o
A. A. A. A. A. A. A. . A. A. A. H A. A. _ A. A. A.
A. A. A. A. A. A. A. Al A. A. , A. A. A. A. A.
- No
A. A. A. A. A. A. A. A. A. A. A. A. A. A. A.
A. A. A. A. A. A. Al A. A. A. A. A. A. A. A.
. - v.0
Al Al A. A. A. A\ A. A. A. A.
Al Al A. A. A. A\ A. A. A. A.
o - 9o
4.. AP A. A. A. A! A A. A. A.
I 4! 4r 4. A. 04. R A. A. A.
i n u n u n u n n n
t f l I. ! I. \ Q t. K .
.. ._ J . ... . ., .. .. .. .. mo
o.—

 

 

 

142

Avd as“: b 980 .Su SE ...—3:8 grab—.850 .36 PEME

as 22523“

®®.N 0m... 00; ®v._ ON; 00.? 0m..® 00.0 OV.® ®N.® ®®.®
_ _ _ _ ._ _ _ _ _

 

®®.®

0N6

®V.®

®o.®

 

 

®m.®

 

 

 

 

®®._

(w) NOILD'ERIICI'A

143

Av.» ...—act b 930 .8.— «o... .883 hag—Q,

.3.» 8%.:

 

 

 

 

 

 

 

ON «A m: o; «.0 o 0 v6 Nd od
.. .. . 1 . . 3
A. A. m A. A. H A. A. Al Al Al A. A. A. A. w A. A.
A. A. ”A. AI A. A| Al Al Al Al A. A. A. V A. A.
. ................. .. . : : ........ .I Nd
A. A. WA. A. MA. A. Al Al A. A! A. A. A. A. A.
A. A. WA. A. .A. A\ Al A! Al Al AI 4. A. A. A.
Al Al A. A. A\ N f A. A. 0
AI 4. A. A. I N A 0 O A.
U o ..... l 0.0
Al 41 WA. A. 2 fl A A A.
W . .
A. A. A. f of A A
4M 4r 4. f 4r fl A A ¢
Ar Ar 4, 4r fl
1 I v A. A. 4' u u “ .
A .. . ._ .. .. .. .. .. mo
A:

 

 

 

144

 

 

 

 

 

 

Variable Value Source
dispersivity, a 0.061 cm Pennel et al. 1993
dispersivity, a (with 0.136 :t 0.021 cm Pennel et al. 1993
trapped residual NAPL) .

Sm 0.179 :t 0.004 measured
density NAPL, p" 0.748 i 0.003 g/cm3 measured
equilibrium solubility of 2005 j: 40 ppm measured (batch
NAPL in 5% surfactant, experiment)

CS

 

lumped mass transfer
coefficient, k,

0.387 a: 0.047 hr -1

measured (flow
interruption study)

 

 

porosity, n

 

0.38 i 0.013

 

measured

 

 

Table 6.1. Soil and fluid properties used in model simulations.

 

 

 

 

 

 

 

 

 

 

Contaminant k, (hr '1) Reference
Naphthalene 0.081 Priddle and MacQuarrie 1994

Acenaphthenc 0.055 "
Flourene 0.074 "
Phenanthree + anthracene 0.079 "

Benzene 0.0001 Borden and Kao 1992
Tolunc 0.0004 "
Xylene 0.0004 "
Aliphatic hydrocarbon 0.0004 "

 

 

 

Table 6.2. Representative range of mass transfer coefficients (k1).

 

145

 

 

 

 

 

 

 

k, (hr ‘1) Sm on (em) Maximum relative error (%)
0.434 0.179 0.136 28%
0.387 0.183 0.136 94%
0.387 0.179 0.157 0.6%
0.387 0.179 0.061 2%
0.434 0.183 0.157 10.3%

 

 

 

 

Table 6.3. Error propagation for the lumped mass transfer coefficient (k1),
saturation of trapped residual NAPL (Sn). and dispersivity (a).

 

146

...—95:3 8.3.6.3 he eases—«>0 o... .8.— 8Eog3 =9: ..e baa—Eam 6.» 03:.

 

:03 Bag :8 on he.“ 59% omega .

290mm 98.58.4528; Bod Eagamgm oz ..
$529. maomzmoozom E Bod Eagéazbm oz ...

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

<2 <2 83 83 N3 2.. ...2 on; ... a
<2 <2 83 83 N3 9: 35 :3 .. a
33 :4 :3 23 m3 32 92 on; a: ...: .3 .3 «2 n3 5
games... 2:33 So 83 N3 :.._ 33 33 33 33 .3 .3 n2 n3 3
:3 t; :3 m3 .3 a; 1.. «S a; 3: m3 m3 n2 «S n
33 33 :3 33 23 33 «3 33 a; a; .3 .3 33 n3 3
n3 :3 23 23 e3 :3 R; s; a: n3 .3 .3 24 2.. m
23 MS :3 :3 W3 32 n2 «2 a; a; .3 .3 n2 ...? N
<2 $2 $3 83 .3 33 83 84 a; a; $3 83 n2 34 a
<2 <2 :3 23 m3 ”2 o: 32 a; a; $3 $3 24 n2 2
No x. 8% «So «So 35 $5 3 5 "m G ..a ”a ..a w ”.65

 

 

Chapter 7
Dissertation Summary and Recommendations

7 .1 Summary

The large number of sites identified for remediation by the United States
Environmental Protection Agency supports the need for engineering solutions to confine
damage, protect drinking water and to decontaminate aquifers. These sites vary widely in
the geological characteristics and types of contaminants present, placing further demands
on the technologies developed. This research presents a feasibility study of the hydraulic
measures which influence the flow pattern and the resulting removal of a nonaqueous
phase source of contamination within the porous matrix. Although focused on the use of
chemically enhanced dissolution of trapped residual LNAPL at the capillary fi'inge,
problems of in-situ delivery and recovery of treatment is common to other remediation
techniques.

The design parameters evaluated were the number of wells, the location of the
well, and the pumping rates. The primary objectives of the use of hydraulic controls were
to: 1) deliver surfactant into the contaminated region; 2) ensure sufficient contact time of
the surfactant with the contaminated region; and 3) to recover contaminant laden
surfactant enhanced ground water. A

The initial three chapters of this dissertation present an introduction to the
research, a literature review and an overview of the materials and methods used in the
experimental work. The latter three chapters contain the results of experimental and
numerical studies to promote surfactant enhanced dissolution of trapped residual NAPL

at the capillary fringe. Each phase of the research has been presented in a separate

147

148

chapter that can stand alone, with a general summary presented here. The reader
interested in more detailed summaries of individual phases of the work is directed to the
summary section at the end of each chapter.

The first experimental phase of the dissertation is presented in Chapter 4. The
specific goals included: 1) developing an analytical technique to . measure the
concentration of dodecane in a surfactant enhanced aqueous phase using head space gas
chromatography; 2) comparing the enhanced solubility of NAPL measured in batch
experiments with that observed when the NAPL is trapped as a residual in a porous
media; and 3) investigating the non-equilibrium effects of surfactant enhanced solubility
of trapped residual NAPL. The use of surfactants caused heavy sudsing and the
formation of an emulsion during both liquid-liquid extraction and solid-liquid extraction
procedures. As a result, an analytical method was developed using head space gas
chromatography using a Perkin Elmer PE Autosystem gas chromatograph equipped with
a flame ionizer detector and a PEHS40 automatic headspace sampler. The use of gas
chromatography to measure the concentration of NAPL in surfactant was shown to be
dependent upon the partitioning of the NAPL from the aqueous phase into the gaseous
phase. Tests show that the degree of partitioning is strongly influenced by surfactant
concentration in the aqueous phase. To overcome the effect of surfactant concentration,
samples are diluted 100:1 in 5% surfactant. The detection limit for the gas
chromatography is approximately 10 ppm.

Batch experiments showed that a 5% solution of Witconol NP-l 50 enhances the
aqueous phase solubility or dodecane fiom 0.035 ppm to approximately 2000 ppm. In
the porous matrix, the solubility is enhanced to approximately 1600 ppm, due in part to
the reduced contact area available when NAPL is trapped as a ganglia compared to the
complete mixing in batch systems. Flow interruption studies were used to measure the

lumped mass transfer . coeflicient of dodecane into a 5% solution of Witconol NP-150

149

which concluded that the dissolution of dodecane into the surfactant is a rate-limited
process, but not prohibitively slow.

Chapter 5 presents the second experimental phase of the study. Column studies
were conducted to develop a better understanding of the relative aqueous phase
permeability in the region near the capillary fiinge. Previous measurements by petroleum
engineers of the relative permeability-saturation relationship of the aqueous phase beyond
the regions where both phases are continuous have been limited to studies in sandstone
and glass beads. This research considered that relationship over a range of trapped
residual immobile phase saturations in a natural unconsolidated soil. As with the previous
research by petroleum engineers, these experiments required a separate column
experiment for each data point. The maximum aqueous phase saturation was reduced to
approximately 85-89% due to the trapped residual nonwetting phase, reducing the
relative permeability of the aqueous phase to 0.55k, - O.8k,.. A separate experiment was
conducted using a single column in a long term study of the enhanced dissolution of
trapped residual NAPL to provide a continuous measurement in this region. The results
show that the continual increase in the aqueous phase saturation and relative permeability
follows the trend of the previous experiment. The combined results of these two
experiments provide a detailed image of the relationship of the relative permeability and
saturation of the aqueous phase in this region. A third set of experiments was also
conducted that measured the relative permeability of the aqueous phase in soil columns
where both a trapped residual NAPL and a trapped residual air phase were present. The
results show that the two trapped residual phases reduce the saturation of the aqueous
phase to approximately 50% - 85%, which is a larger range that with a single phase
trapped. However, the aqueous phase permeability remains at a relatively constant value
of approximately 0.351., compared to the larger range when only one phase is trapped as
a residual. Fitting parameters derived from curve fitting pressure-saturation curves in

research conducted by Soeryantono et al. (1994) and Powers et al. (1992) were utilized

150

to evaluate the use of Corey's and van Genuchten's equations for the permeability-
saturation relationship in the region where the aqueous phase permeability is increased as
the trapped residual nonwetting fluid is decreased. The results indicate that the fitting
parameters do not adequately predict the relationship in this" region of interest.
Therefore, the raw data may need to be used directly in design and modeling.

An implicit l-D transport code was developed in Chapter 6 to predict the removal
of trapped residual NAPL using the lumped mass transfer coefficient measured in Chapter
4. Results were presented using dimensionless parameters which represent velocity,
length, mass transfer and dispersion. In cases where horizontal flow may be assumed in
an aquifer, the model is usefirl in optimizing the pumping rate when considering the
number of pore volumes of surfactant required for treatment, the amount of waste water
generated and the residence time required to promote near equilibrium saturation of
dodecane into a surfactant enhanced aqueous phase. A numerical simulation was
compared against actual data collected from the effluent end of a column flushed
continuously with surfactant. The model did not predict the trend in the experimental
data at later time which can be explained in part by the limitations in the use of the
lumped mass transfer coefficient. This coefficient is a fimction of both the specific
interfacial area between the NAPL and the aqueous phase and a measure of the resistance
to mass transfer across this interface. Not only is the actual internal contact between the
aqueous phase and the NAPL difficult to quantify experimentally or mathematically due
to the complex size and shapes of the ganglia, but the interfacial surface area changes
with time as the NAPL dissolves into the aqueous phase. A second limitation of the
model may exist if preferential flow paths or pore constrictions are present within the soil
prior to or during treatment of the contaminated soil with surfactant. This in part also
explains why the enhanced solubility of NAPL is reduced in the porous matrix compared
to the enhanced solubility achieved in batch experiments.

151

An existing USGS flow and transport model for unsaturated-saturated flow was
used to evaluate the hydraulic controls a 2-D laboratory scale model aquifer in Chapter 6.
Pumping rates, the number of wells, and the location of wells were evaluated using
SUTRA in addition to the 1-D implicit transport model which evaluated the residence or
contact time directly. Delivery of the surfactant enhanced aqueous phase and the
recovery of the contaminated fluid was determined from velocity vector plots and
surfactant concentration contour plots which predicted the flow paths resulting from the
hydraulic controls. The study Was limited to a 2 m x 1 m vertical cross section of an
unconfined sandy homogeneous aquifer, with a region of contamination near the capillary
fiinge. This dimension was chosen to represent an existing laboratory model at Michigan
State University. However, the numerical prototype can easily be modified to further this
research. The contaminated region was modeled as a heterogeneous region in the aquifer
due to the reduced relative permeability of the aqueous phase resulting from the trapped
immobile immiscible phase. Values of the aqueous phase permeability were selected from
the measured values in the ID column studies. Model simulation of a vertical cross
section indicate that the reduced relative permeability in this region does not cause major
by-passing. The optimum delivery and recovery of the surfactant was achieved when
‘maintaining local flow velocities near the range of ambient flow velocities prior to
pumping. These low pumping rates maintained relatively horizontal flow paths in the
aquifer. As a result, the 1-D model was used to evaluate the residence time to predict
whether sufficient contact time was achieved. The ID and 2-D modeling indicate that
surfactant can be delivered throughout the contaminated region and recovered from the
aquifer using hydraulic controls which also promote sufficient contact time with the
contaminated region.

Through this research effort, an understanding of the interactive roles of rate-
limited mass transfer in remediation, reduced relative permeability near the capillary

fiinge, and the use of hydraulic controls near the capillary fiinge has been advanced.

152

Experimental methods and an implicit l-D numerical transport model were developed and
used in conjunction with an existing 2-D code to evaluate the in-situ use of surfactant
enhanced remediation of trapped residual NAPL near the capillary fiinge. This work is of

interest to researchers in immiscible flow in porous media and in-situ aquifer remediation.

7.2. Summary of Costs

This section is provided to give a general overview of the costs associated with
the in-situ use of surfactant enhanced aquifer remediation. Roy F. Weston, Inc. (1988)
outline additional design considerations, equipment requirements, treatment needs,
disposal needs, monitoring requirements, pemiitting requirements, exposure pathways,
environmental effectiveness and costs involved with in-situ leaching, which includes
surfactant enhanced aquifer remediation, based on surveying past remediation. In the
survey, the capital cost data is scarce, due to limited experience with these methods of
remediation. As a result, the costs are limited to equipment needs which are summarized
in Table 7 .1. The cost of surfactant (Witconol NP-150) is approximately $0.99 per
pound. For a 5% solution of surfactant, the treatment costs are approximately $1 lO/m3.
For the contaminated region evaluated (0.8 m x 0.3 m), one pore volume of treatment

costs approximately $106 per unit width.

7 .3 Recommendations

The following list overviews possible directions for future research based on the

dissertation presented:

. Investigate the lumped mass transfer coefficient for varying surfactant
concentrations.

. Investigate the use of SUTRA to predict the removal of an immobile phase while

monitoring the concentration of the immobile phase.

153

. Develop or acquire a model that includes the transport of the miscible surfactant
and the transport of an immiscible immobile phase in the surfactant enhanced
aqueous phase.

. Conduct a similar 2-D study for a less permeable soil to consider hydraulic
controls in a region where slow velocities limit the optimum use of the surfactant
to enhance the dissolution of trapped residual NAPL.

. Extend the model aquifer region to include both a larger cross sectional area of
the aquifer in addition to a larger contaminated region.

. Investigate the effects of local heterogeneities caused by layers of clay and silt
with respect to this study.

. Conduct an experiment in the 2-D laboratory scale model to either qualitatively or
quantitatively measure the size and shape of the discontinuous NAPL ganglia. In

addition, measure the hydraulic conductivity in the 2-D setting.

Current research is focused on predicting the mass transfer rate of NAPL into the
aqueous phase based on measurable fluid or soil properties which may predict the
changing size and shape of the residual NAPL blob. A study which evaluated the effect
of surfactant concentration on the lumped mass transfer coefficient may contribute to the
understanding of the dissolution process. SUTRA can include an immiscible
contaminant, but does not track the changing volume of the immiscible phase in the
output. In addition, it can not model both the transport of the surfactant enhanced -
aqueous phase and the transport of the solubilized NAPL. A model may be available with
these capabilities. The study should be expanded to investigate hydraulic controls in
aquifers where the residence time is difficult to achieve. In a less permeable soil, the
trapped residual NAPL may have a more pronounced effect on the flow within the
aquifer, causing the surfactant enhanced aqueous phase within the contaminated region to

solubilize dodecane at equilibrium solubility concentration during the initial migration of

154

the treatment into the region. Using the same aquifer material, a larger aquifer region
should be studied based on similar reasoning. For the aquifer soil studied, the relative
permeability of the soil is not reduced by the presence of trapped residual oil as
significantly as localized clay and 'silt layers which should also be considered in a more
realistic representation of in-situ environment. Past research in identifying ganglia size
and shape has been in 1-D columns, where the column walls may limit the size of the
ganglia, as suggested by the stable ganglia length calculations. The laboratory scale
physical model could be used to measure ganglia in addition to measuring the hydraulic

conductivity of the aquifer material and in the region of contamination.

 

 

 

 

 

 

 

Item Description ‘ Approximate
Cost“
Capital Costs Submersible pumps (wells) 50-2000
Centrifugal pump (chemical 8400-81000
feed)
PVC pipe 0.15 m diameter $100-$600
Installation Costs PVC pipe $49.20-$65.60/m $65.6-$82/m
Operation and Sampling and analytical costs 4 samples per year, 3 $49.2-865.6/m
Maintenance monitoring wells
Costs

 

 

 

 

Table 7.1. Summary of costs in 1986 dollars involved with in-situ leaching (Roy F.
Weston, Inc. 1988). ‘

 

Appendices

Appendix A

Analytical Solution to the 1-D Transport Equation
(van Genuchten and Alves 1982)

 

6C 52C 6C
R——=D — —— C+
a. 3x2 vé’x p y
Initial and Boundary Conditions
C(x,0)=C,-
C 0 t t
C(O,t)={ ° < < °
0 (>10
6C
,1 =0
§_x(oo )
1+[Q—Z)A(x,t)+[C-(x,—Z)Bt) 0<t<to
C(x,t)=(p .11 #
1+[Q—Z]A(xt)o+(C-xZ-)H()‘C3(x"’o) t>to
_# ll .11

 

where:

Rx— vt 1 Rx+vt
A(x, t): exp(- %t){l——etfn{2(0 INN-Sap )r2(DRt)i]

B(x,=t) %exp[(u2v)x)] ]e Rx 2(DRm]+ +exp-;-[ [(u+v)x)]erf R2;3t,]

 

 

 

 

 

 

i
u: v[l+ 44120)
v

155

Appendix B

Implicit l-D Transport Code for Surfactant Enhanced Dissolution of an
Irnmobile Residual NAPL

PROGRAM MPLICIT

tittttIfitttitittiiiiitfittt$ititi$#***********##*********************¥

LIZE'ITE R. CHEVALIER
MICHIGAN STATE UNIVERSITY
DEPARTMENT OF CIVIL AND ENVIRONMENTAL ENGINEERING

1994
l-D TRANSPORT CODE FOR SURFACT ANT ENHANCED DISSOLUTION

OF A UNIFORM SATURATION OF TRAPPED RESIDUAL NAPL IN
SATURATED SOIL

*§§**§§**§§

tttittfitttttfi¥t#*##*##¢#***##tttttt**ttittt******###**tttittttttttttt

*ttttttfittttttttttltttttttfi¢#***t***t*tttt00*#tttttttttttttttlttttfifit

IMPLICIT CODE

CENTERED DIFFERENCE FOR SPATIAL TERM
FORWARD DIFFERENCE FOR TEMPORAL TERM
TRIDIAGONAL MATRIX

fifl'fittl'l‘l'l

t*ttttttitttitt¥*tifittfitttt****#*#ttttittitttt¥t¥tt¥¢tttttttt*ttttfitt

IMPLICIT DOUBLE PRECISION (A-H,O-Z)

DIMENSION CANAL(5000),C(5000),E(5000),F(5000),G(5000)
DIMENSION X(5000),R(5000)

CHARACTERMO FILEl
COMMON/CCN,RR,D,DX,RKL,CE,DT

MARRAY=5000

fltitittttiltittttittttttttttttfitt##ttitttittitttfiltfiltltttfitttttttfitt
t

"' C-G-S UNITS

‘ .

"' AREA = CROSS SECTIONAL AREA OF COLUMN
* Q = FIJIJX

156

§§§§§§§§§Q§l>§

157

= VELOCITY

= POROSITY
C(X) = CONCENTRATION OF NAPL IN AQUEOUS PHASE AT TIME T
CA(X) = ANALYTICAL SOLN. OF NAPL CONC. IN AQUEOUS PHASE

= MASS TRANSFER RATE COEFFICIENT
CE = EQUILIBRIUM CONCENTRATION OF NAPL IN AQUEOUS PHASE
GAMMA = DISPERSIVITY
= DISPERSION

= RETARDATION
COUR = COURANT NUMBER
DA = DAMKOHLER NUMBER

RLENGTH = COLUMN LENGTH

itit.t*ttttttitttttitttttfittttttttitittttt##tfit**¢*****¢*¥$¢t¥*¢¢tt

it.

it.

PRINT *3 PROGRAM INITIATED'

PRINT *: ENTER OUTPUT FILE NAME'

READ(*,'(A40)°)FILE1

PRINT *,' ENTER (1) IF EVALUATING TOTAL POROSITY‘ ‘

PRINT *,' ENTER (2) IF EVALUATING EFFECTIVE POROSITY‘ READ(‘,*)LFLAG
PRINT *3 ENTER ALPHA'

READ(",")ALPHA

PRINT *: ENTER COURANT NUMBER'

READ(*,*) COUR

PRINT #3 ENTER Dx (cm)'

READ("',‘)DX

PRINT ‘3 ENTER (1) IF DISPERSION INCLUDED IN NUMERICAL SOLN
PRINT *,° ENTER (2) IF DISPERSION NOT INCLUDED

READ(",") IFLAG

PRINT I: ENTER DAMKOHLER NUMBER'

READ(‘,*) DA

PRINT t: ENTER LENGTH‘

READ(","‘) RLENGTH

DATA (ml,g,cm)

PRINT ‘,' ENTER ITIME'
READ(*,"')ITIME I
CO=0.0
AREA=23.24D0
DENSITYW=0.998D0
DENSITY O=0.75D0
RN=O.38DO
RKL=0.387/60/60
CE=2000"0.7996
SRNI=0.18D0

RR=1

USER DEFINED DATA

D=V‘0.136
IF(LFLAG.EQ.2)RN=RN‘(l-SRNI)
V=RKL‘RLENGTH/DA
Q=V*RN*AREA

tit

10

23

158

QMIN=Q*6O
DA=RKL*RLENGTH/V
D=V‘O. 136
DT=COUR*DX/V
NCELLS=RLENGTHIDX
BETA=I-ALPHA

SET INITIAL DISTRIBUTION

DO 10 I=l,NCELLS

C(I)=0.0DO
IFCIFLAG.EQ.2)D=O

Al=-2"‘ALPHA"'D*DT-V"ALPHA*DX*DT
A2=2*DX’DX+4*ALPHA*D’DT+2“'ALPHA*RKL‘DX*DX*DT
A3=2*ALPHA"D‘DT+V*ALPHA"DX‘DT
Bl=2"RKL‘CE*DX*DX‘DT
BZ=2‘BETA‘D‘DT+V*BETA*DX*DT
B3=-2’RKL‘BETA*DX"DX*DT+2*DX‘DX-MBETA‘D‘DT
B4=2*D‘BETA‘DT-V‘BETA‘DT‘DX

D=V‘0. l 36

OPEN(5,FILE=FILE l)
WRITE(5, 101)
WRITE(5,102)DA,COUR
WRITE(5, 104)V,RLENGTH
WRITE(5,105)DX,DT
WRITE(5, 106)NCELLS,QMIN
WRITE(5, 103)ALPHA

IF (IFLAG.EQ.2) WRITE(5,110)
DO 20 J=l,ITIME

DO 23 I=l,NCELLS
CALL ASOLN(CA,I,I)
CANAL(I)=CA

IF(I.EQ.1) THEN

F(I)=A2

G(I)=A3 R(I)=Bl+C(I)‘B3+C(I+l)"'B4-Al*CO GOTO 23
ENDIF

IF(I.EQ.NCELLS)THEN

E(I)=Al

F(I)=A2+A3 R(I)=Bl+C(I-l)*BZ+C(I)*(B3+B4) GOTO 23
ENDIF

E(I)=AI

F(I)=A2

G(I)=A3 R(I)=Bl+C(I-l)"‘BZ+C(I)*B3+C(I+1)*B4
CONTINUE

CALL TRI(E,F,G,X,R,NCELLS)

25

20

28

101

102
103
104
105
106
112
110
120

159

D0 25 I=l,NCELLS
C(I)=X(D

C(NCELLS+1)=C(NCELLS)

CONTINUE

TTIME=J*DT/60/60/24
WRITE(5,1 12)TI'IME

DO 28 I=l,NCELLS XLOC=(DX*(I-l)+0.5*DX)/RLENGTH CA=CANAL(I)/CE
CI=X(I)/CE

ERROR=CANAL(I)-X(I)
PER=ABS(ERROR/CANAL(I)* 100)
NCOUNT=I+NCOUNT

IF (NCELLSGT. 100) THEN

IF(NCOUNT.EQ.10) THEN
WRITE(5,120)XLOC,CA,CI,ERROR,PER
NCOUNT=0

ENDIF

ELSE

WRITE(S, 120)XLOC,CA,C1,ERROR,PER ENDIF
CONTINUE

FORMATC COMPARING ANALYTICAL NUMERICAL SOLUTIONS',/,
+ ' IMPLICIT METHOD',/)

FORMATc DA =',F6.3,' COUR =',F6.3)

FORMATC ALPHA =',F6.3)

FORMATc V(cm/sec) = ',F8.5,3X,' LENGTH (cm) = ',F8.1)

FORMATC DELTA x (cm) = ', F8.3,3X,' DELTA T (sec) = ',F8.2)
FORMATc NUMBER OF CELLS =’,I4,' Q (ml/min) = ',F8.3)

FORMATC TOTAL TIME (days) ',F8.3//‘X/L CA C ERROR %°,/)
FORMATc DISPERSION NOT IN NUMERICAL SOLN.')
FORMAT(FS.3,2(1X,F5.3),1X,F8.3,1X,F10.3)

PRINT *,' PROGRAM TERMINATED NORMALLY‘

1000 END

ttitttttitttfititttiittt*ttttittttfitt******¢***#***********t**#*#*¥*

SUBROUTINE ASOLN(CA,I,J)
IMPLICIT DOUBLE PRECISION (A-H,O-Z)
COMMON/CC/V,RR,D,DX,RKL,CE,DT

GAMMA=RKUCE
TERM=GAMMA/RKL

R=RR

X=I*DX-0.5*DX

T=J*DT

CI=0.0D0

CO=0.0D0
DENOM=2*(D"'R*T)"*O.5
U=V‘(1+4*RKL*D/V“2)“O.5

160

AA4= l-ERF((R"X +V*T)/DENOM)

IF(AA4.EQ.0)THEN

A=DEXP(-RKL*T/R)*(1-0.5"'(l-ERF((R"‘X-V*T)/DENOM)))

ELSE

=DEXP(-RKL‘T/R)*(l-O.5*(l-ERF((R*X-V*T)/DENOM))

+ -0.5*DEXP(V"‘X/D)"(l-ERF((R*X +V*”D/DENOM)))

ENDIF

BB4=l-ERF((R*X+U"‘T)/DENOM)

IF(BB4.EQ.0)THEN

B=0.5*DEXP((V-U)*X/2/D)*(I-ERF((R*x-U*T)/DENOM))

ELSE

B=O.5‘DEXP((V-U)*X/2/D)*(l-ERF((R*X-U"T)/DENOM))+
+ 0.5*DEXP((V+U)*X/2/D)*(l-ERF((R*X+U*T)/DENOM))

ENDIF

CA=TERM+(CI-TERM)*A+(CO-TERM)*B

RETURN

END

iitt***¥*¥#*tttiittfifitttt*********¥*¥****#it***##**************#**¥

DOUBLE PRECISION FUNCTION ERF(X)
IMPLICIT DOUBLE PRECISION (A—H,O-Z)
IF (X.EQ.0.0) THEN
ERF=0.0
RETURN
ENDIF
IF (X.LT.0.0) THEN
I=I
x=-x
ELSE
I=0
ENDIF
P=0.32759ll
AI=0.254829592
A2=-0.284496736
A3=l.42141374l
A4=-1.453152027
A5=1.061405429
T=l.O/(l.0+P"'X) ERF=1.o-(A1*T+A2*T**2+A3*T**3+A4*T**4+A5*T"5)*
+ DEXP(-X*X)
IF(I.EQ.1) THEN
ERF=-ERF
=-x
ENDIF

RETURN

END

##ttfittttififitt¥t#¢#¢¢*##**#*¥*tttt*#**#*************$¥¢**¥t*$$$*t*¥

tit

10

it?

20

##t

30

161

SUBROUTINE TRI(E,F,G,X,R,N)

IMPLICIT DOUBLE PRECISION (A-H,O-Z)
DIMENSION E(5000),F(5000),G(5000),X(5000),R(5000)
DECOMPOSITION

DO 10 K=2,N

B(KFE(K)/F(K-1) F(K)=F(K)-E(K)*G(K-1)

CONTINUE

FORWARD SUBSTITUTION

DO 20 K=2,N
R(K)=R(K)-E(K)*R(K-1)

BACKWARD SUBSTITUTION

X(N)=R(N)/F (N)
DO 30 K=N-l,1,-l

X(K)=(R(K)-G(K)*X(K+1))/F (K
CONTINUE -
RETURN

END

Appendix C

Determination of Mass Transfer Coefficient-Output from
SCIENTIST Sofiware

The following tables were generated from the output of the software used to

determine the lumped mass transfer coefficient, Scientist by MicroMath Scientific

Sofiware (Salt Lake City, Utah).

Table Cl.l. Output from Scientist Software used to determine mass transfer
coefficient, k. = 0.387 hr'l.

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

Goodness-of-fit statistics for data set Model A=A0(l-e'kt)
Weighted Unweighted
Sum of squared observations 11.258 11.258
Sum of square deviations 0.086 0.086
Standard deviation of data 0.064 0.064
R-sguared 0.992 0.992
Coefficient of determination 0.965 0.965
Confidence Intervals ’
Parameter name A0
Estimated value 0.7996
Standard Deviation 0.019
95% Range (Univar) 0.761 0.838
95% Range (S-plane) 0.751 0.848
Parameter Name K
Estimated value 0.387
Standard Deviation 0.047
95% Range (Univar) 0.290
95% Range (S-plane) 0.264

 

 

 

 

162

163

Table Cl.l- continued.

Variance-Covariance Matrix
0.084
-0.1 193

Correlation Matrix
1 .00
-0.565

 

Information from Residual Analysis:

The following are normalized parameters with an expected value of 0.0

Values are in units of standard deviaton form the expected value.

The serial correlation is 1.71 which indicates a systematic non-random trend in the
residuals.

Skewness is -1.58 indicating the likelihood of a few large negative residuals having an
unduly large effect on the fit.

Kurtosis is 0.06 which is probably not Significant.

The weighting factor was 0.00 leading to a heteroscedacticity of -0.54 which suggests an

optimal weight factor for this fit of ab0ut -O.54.

164

Table C1.2. Output from Scientist Software used to determine mass transfer
coefficient, k. = 0.161 hr'l.

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

Goodness-of-fit statistics for data set Model A=(1-e'kt)
Weighted Unweighted
Sum of squared observations 11.258 11.258
Sum of square deviations 0.307 0.307
Standard deviation of data 0.118 0.118
R-squared 0.972 0.972
Coefficient of determination 0.697 0.697
Correlation ~ 0.962 0.962
Confidence Intervals
Parameter name K
Estimated value 0.161
Standard Deviation 0.015
95% Range (Univar) 0.129 0.193
95% Range (S-plane) 0.129 0.193
Variance—Covariance Matrix:
0.017
Correlation Matrix:
1.00

 

 

 

 

Information fi'om Residual Analysis:

The following are normalized parameters with an expected value of 0.0

Values are in units of standard deviaton form the expected value.

The serial correlation is 3.33 which indicates a systematic non-random trend in the
residuals.

Skewness is -1.77 whichis probably not significant.

Kurtosis is 0.08 which is probably not Significant.

The weighting factor was 0.00 leading to a heteroscedacticity of -1.25 which suggests an

optimal weight factor for this fit of about -1.25.

Appendix D
Summary of Input Data for SUTRA Model

165

 

3 we.» as. 225 B SEES

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

8 8 5mm

.. .. 8% 85 E a 8323 5832 V22:

_ _ 85:83; @0580 3:20

85 3o eou§8§32§ .5sz

no 3 mam»? 888% 3885 .5

2: 8" :2 389223 B 88 E can: 5 2% 3 83> 96.5.

...... a... g 7 gm 200 ~+ amm—

c o ugh 3385 o 3% 88am _+ Emm—

_ _ >55 E225 o 2% 38a I 0.59

c _ 8.238 c 3839.5 1 e322

o c 893.88 3.. 89... Saw E .36.... 5252 82.2

c c 88.. 8:388 E .352 momz

a c 3839. was 228 55 88.. B 5:32 . . :82

m e E3038 55 88.. E .352 momz

o o .88 858% a? 88: B .352 252

a an 238a Reece. a? 88: B .8852 232

o c 5252

a :.
8N 8m 38%.» B .3552
5 an 88.. B exaaz
EOE: 22> meeam
.h—fio Egan .........,.._....,..w..u_a...»....H...,.....,..4.3m..m”..H.n.....n...._.,........«A»,an..h...n..,.u...a..uflmm..».fiwuw.flis.”...“.w.........“CHAIRMAN.,. : . . ...H.” ...... . ..........
Ban—Em 3838:: w... .

 

 

  

 

 

22335 3582. SEE E 8...: 3.5

382 gm E 98 sea .8 3585

a xenon?

AAQ might

 

 

 

dos—5:8 - .80ng 10:08:: gm 5 com: San— AAD mama.

167

 

gigguvumaduofificuogagg a
Sggggggo§> 9..

don—.mbmmD SE @023

dos—5:8 - macaw—25m flow—0:5: gm 5 com: San— ._ . a mama?

Appendix E
Materials and Suppliers

Ottawa Sand

Soiltest

86 Albrech Dr.

PO. 8004

Lake Bluff, Illinois 50044-8004

Telephone: 1.800.323. 1242

Description: Ottawa density sand; p approximately 93 lb/ft3 (1.49 g/cm3)

Dodecane

Phillips 66 Company

A Subsidary of Phillips Petroleum
Box 968

Borger,Texas 79008-0968
Telephone: 806.274.5236
Description: Technical grade

Witconol NP-l 50

WITCO

1 American Lane

Greenwich, Connecticutt 06831
203.552.2000

Description: Product Code 087-5935

Oil Red 0 Biological Stain
Aldrich Chemical Company

PO. Box 355

Milwaukee, Msconsin 53201-9358
Telephone: 1.800.558.9160
Description: Stock # 19,819-6

168

List of References

List of References

Abdul, AS, Gibson, T.L., Ang, C.C., Smith, and J.C. Sobczynski. 1992. ”In-Situ
Surfactant Washing of Polychlorinated Biphenyls and Oils from a Contaminated
Site”, Ground Water, 30(2): 219-231.

Abriola, L.M. 1989. "Modeling Multiphase Migration of Organic Chemicals in
Groundwater Systems - A Review and Assessment”, Environmental Health
Perspectives, 83:117-143.

Amaefule, J .O. and LL. Handy. 1981. ”The Effect of Interfacial Tensions on Relative
Oil-Water Permeabilities of Consolidated Porous Media", SPE/DOE 9783, presented
at 1981 SPE/DOE Second Joint Symposium on Enhanced Oil Recovery, Tulsa, 0k,
April 5-8.

American Petroleum Institute. 1972. ”The Migration of Petroleum Products in Soil and
Ground Water: Principles and Countermeasures", API Publication No. 4149.

Anderson, R.L. 1987. Practical Statistics for Analytical Chemists, Van Nostrand
Reinhold, pp.94-107.

Anderson, MP. 1979. "Using Models to Simulate the Movement of Contaminants
Through Groundwater Flow Systems", CRC Critical Reviews in Environmental
Control, 9(2): 97-156.

Anderson, MP. and W.W. Woessner. 1992. Applied Groundwater Modeling: Simulation
of F low and Advective Transport, Academic Press.

Ang, C.C. and AS. Abdul. 1991. "Aqueous Surfactant Washing of Residual Oil
Contamination fi'om Sandy Soil”, Spring Gm, pp. 121-127.

Battermann, G. 1986. "Decontamination of Polluted Aquifers by Biodegradation",
Contaminated Soils, First International 7N0 Conference on Contaminated Soil,
Edited by J. S. Assink and WJ. van den Brink, Martinus Nrjhofl‘ Publishers,
Dordrecht, Netherlands.

Bear, I. 1979. Hycb'aulics of Grounaivater, McGraw-Hill, New York, NY.

169

170

Bear, 1., Beijin, MS. and RR Ross. 1992. ”Fundamentals of Ground Water Modelling”,
EPA/540/S-92/005.

Berg, RR 1975. ”Capillary Pressures in Stratigraphic Traps”, AAPG Bull, 59(6):939-
956.

Borden, RC. and C. Kao,. 1989. ”Water Flushing of Trapped Residual Hydrocarbon:
Mathematical Model Development and Laboratory Validation“, Proceedings of the
Conference on Petroleum Hydrocarbons and Organic Chemicals in Ground Water:
Prevention, Detection and Restoration, NWWA, API, pp. 473-486.

Chapra, SC and RP. Canale. 1988. Numerical Methods for Engineers, 2nd Edition,
McGraw-Hill,.

Chatzis, I., Morrow, NR, and HT. Lim. 1983. "Magnitude and Detailed Structure of
Residual Oil Saturation”, SPEI, 23:311-326.

Chatzis, I. and NR. Morrow. 1984. ”Correlation of Capillary Number Relationships for
Sandstone”, SPEJ, 24(5):555-562.

Chatzis, I, Kuntamukkula, MS, and NR Morrow. 1988. ”Effect of Capillary Number
on the Microstructure or Residual Oil in Strongly Water-Wet Sandstones", SPE
Reservoir Eng. , 3(3):902-912.

Chatzis I. and PAL. Dullien. 1983. "Dynamic Immiscible Displacement Mechanisms in
Pore Doublets: Theory Versus Experiment", J. Coll. Inter]? Sci, 91(1): 199-222.

Clarke, AN, Mutch Jr., RD., Wilson, D.J., and KR Oma. 1992. ”Design and
Implementation of Pilot Scale Surfactant Washing/Flushing Technologies Including
Surfactant Reuse", Wat. Sci. Tech, 26(1-2):127-l35.

Clarke, A.N., Plumb, P.D., Subramanyan, T.K., and D.J.\Vrlson. 1991. ”Soil Clean-Up by
Surfactant Washing. 1. Laboratory Results and Mathematical Modeling”, Sep. Sci.
Technol, 26(3): 301-343.

Conrad, S.H., Wilson, J.L., Mason, WR. and WJ. Peplinski. 1992. ”Visualization of
Residual Organic Liquid Trapped in Aquifers", Water Resour. Res, 28(1): 467-478.

Conrad, SH, Hagan, ER, and J.L. Wilson. 1987. ”Why are Residual Saturations of
Organic Liquids Different Above and Below the Water Table?”, Presented at
NWWA/API Conference on Petroleum Hydrocarbons and Organic Chemicals in
Ground Water - Prevention, Detection and Restoration.

Corey, AT. 1984. Mechanics of Immiscible Fluids in Porous Media, Water Resources
Publication.

171

Craig, F.F., Jr. 1971 The Reservoir Engineering Aspect of Waterflooding, Henry L.
Hoherty Series: Monograph Volume 3, Society of Petroleum Engineers of American
Institute of Mining, Metallurgical and Petroleum Engineers, N.Y.

Edwards, DA. and RG. Luthy. 1990. "Nonionic Surfactant Solubilization of Polycych
Aromatic Hydrocarbons in Aqueous and Soil/Water Systems”, Environmental
Engineering, Proc. of the 1990 Specialty Conference, C.R. O'Melia, Ed., Arlington,
Va. pp. 286-289.

Ellis, W.D., Payne , JR, Tatum, AN., and F]. Firestone. 1984. "The Development of
Chemical Countermeasures for Hazardous Waste Contaminated Soils",
EPA/600/2-85/129.

Ellis, W.D.,Payne, J.R., and GD. McNabb. 1985. ”Treatment of Contaminated Soils
with Aqueous Surfactants”, EPA/600/2-85/129, NTIS PB 86-122 561/REB.

Fischer, H.B., List, E.J., Koh, RC.Y., Imberger, J., and NH. Brooks. 1979. Mixing in
Inland and Coastal Waters, Academic Press.

Fountain, J.C. 1991. "Field Tests of Surfactant Flooding: Mobility Control of
Nonaqueous-Phase Liquids”, Transport and Remediation of Subsurface
Contaminants-Colloidal, Interfacial and Suflactant Phenomena, Edited by DA
Sabatini and RC. Knox, pp. 182-191.

Freeze, R. A. and IA. Cherry. 1979. Groundwater, Prentis Hall, Inc., Englewood Cliffs,
NJ.

Gilliland, HE. and FR. Conley. 1975. ”Surfactant Waterflooding", Proc. of the 9th
World Petroleum Congress, Tokyo, Japan, May 11-16.

Grubb, S. 1993. ”Analytical Model for Estimation of Steady-State Capture Zones of
Pumping Wells in Confined and Unconfined Aquifers”, Ground Water, 31(1):27-32.

Hall, CW. 1989. ”Practical Limits to Pump and Treat Technology for Aquifer
Remediation”, Proceedings of Conference on Prevention and Treatment of Ground
Water.

Honarpour, M. Koederitz, L. and AH. Harvey. 1986. Relative Permeability of
Petroleum Reservoirs, CRC Press.

Hunt, JR, Sitar, N., and K.S.Udell. 1988. "Nonaqueous Phase Liquid Transport and
Cleanup: 1. Analysis OfMechanisms”, Water Resour. Res, 24(8):]247-1258.

Johnson, RE., and RH. Dettre. 1966. ”The Wettability of Low-Energy Liquid Surfaces",
J. Coll. Interf Sci, 21, 610-622.

172

Keely, J .F. 1989. "Performance Evaluations of Pump and Treat Remediations",
EPA/540/4-89/005.

Klute, A. 1986. Methods of Soil Analysis, Part 1. Physical and Mineralogical Methods,
American Society of Agronomy, Inc., Soil Science Society of American, Inc.,
Second Edition, Madison, WI.

Knox, RC. and DA. Sabatini. 1991. "Transport and Remediation of Subsurface
Contaminants”, Transport and Remediation of Subsurface Contaminants-Colloidal,
Interfacial and Surfactant Phenomena, Edited by DA Sabatini and RC. Knox, pp.
1-1 1. _

Kueper, RH. and ED. Frind. 1988. "An Overview of Immiscible Fingering in Porous
Media”, J. Contam. HydroL, 2: 95-110.

Larson, RG., Davis, HT, and LE Scriven. 1981. ”Displacement of Residual
Nonwetting Fluid fi'om Porous Media”, Chem. Eng. Sci, 36: 75-85.

Levich, V.G. 1962. Physiochemical Hydrodynamics, Prentice Hall.

Mace, RE. and J .L. Wilson. 1992. "Clay and Immiscible Organic Liquids: Greater
Capillary Trapping of the Organic Phase”, Transport and Remediation of Subsurface
Contaminants: Colloidal, Interfacial and Surfactant Phenomena, Edited by DA.
Sabatini and RC. Knox, ACS Symposium Series No. 491:205-216.

Mackay, D.M. and J .A Cherry. 1989. "Groundwater Contamination: Pump-and-treat
Remediation”, Environ. Sci. T echnol. , 23 (6), 630-636.

Mackay, D.M., Roberts, P.V., and J.A. Cherry. 1985. ”Transport of Organic
Contaminants in Groundwater”, Environ Sci. Technol., 19(5):384-392.

Melrose, J .C. and CF. Brandner. 1974. ”Role of Capillary Forces in Determining
MicrOSCOpic Displacement Efficiency for Oil Recovery by Waterflooding", JCPT,
13(4): 54-61.

Mercer, J .W. and RM. Cohen. 1990. "A Review of Immiscible Fluids in the Subsurface:
Properties, Models, Characterization and Remediation”, J. Contain. Hydrol, 6: 107-
163.

Miller, C.T., Poirier-McNeill, MM, and AS. Mayer. 1990. "Dissolution of Trapped
Nonaqueous Phase Liquids: Mass Transfer Characteristics”, Water Resourc. Res.,
26(1 1):2783-2796.

Mohanty, K.K., Davis, HT, and LE. Scriven. 1981. ”Thin Films and Fluid Distribution
in Porous Media”, Surface Phenomena in Enhanced Oil Recovery, Edited by DC.
Shah, Plenum Press, pp. 595-609.

173

Morrow, NR and B. Songkran. 1981. "Effects of Trapping and Buoyancy Forces on
Non-Wetting Phase Trapping in Porous Media”, Surface Phenomena in Enhanced
Oil Recovery, Edited by DD. Shah, Plenum Press, pp. 387411.

Morrow, NR, Chatzis, I., and H. Lim. 1985. "Relative Permeabilities at Reduced
Residual Saturation”, JCPT, July-August, 24(4): 62-69.

Morrow, NR, Chatzis, 1., and I]. Taber. 1988. ”Entrapment and Mobilization of
Residual Oil in Bead Packs”, SPE Reservoir Eng, 3(3):927-912.

Nash, J.H. and RP.Traver. 1984. ”Field Evaluation of In-Situ Washing of Contaminated
Soils with Water/Surfactants", Proceedings of the Hazardous Material Spills
Conference: Prevention, Behavior, Control and Cleanup of Spills and Waste Sites,
April 9-12, Nashville, Tennessee.

Nash, J.H. 1985. ”Field Studies of In-Situ Soil Washing”, Mason and Hanger - Silas
Mason Company, EPA Contract No. 68-03-3203.

Nash, J.H.. 1988. "Field Studies of In-Situ Soil Washing", EPA/600/SZ-87/110.

Ng, K.M., Davis, HT and LE. Scriven. 1978. ”Visualization of Blob Mechanics in Flow
Through Porous Media", Chemical Engr. Science, 33: 1009-1017.

Palmer, CD. and W. Fish. 1992. ”Chemical Enhancements to Pump and Treat
Remediation”, EPA/540/S-92/001.

Parker, J .C. 1989. ”Multiphase Flow and Transport in Porous Meda", Reviews of
Geophysics, 27(3):31 1-3 28.

Pennell, K.D., Abriola, L.M., Dekker, T.J., and W.J. Weber. 1992. ”Surfactant Enhanced
Solubilization of Entrapped Dodecane in Soil Columns”, American Chemical Society
Division of Environmental Chemistry 203rd ACS National Meeting, San Francisco,
32(1):945-947.

Pennell, K.D., Abriola, L.M., and W.J. Weber. 1993. ”Surfactant Enhanced Solubilization
of Residual Dodecane in Soil Columns: 1. Experimental Investigation”, Environ. Sci.
Technol., 27(12):2332-2339.

Potter, MC. and DC. Wiggert. 1991. Mechanics of Fluids, Prentice Hall, Englewoord
Clifl‘s, NJ.

Powers, S.E., Abriola, L.M., and W.J. Weber. 1992. ”An Experimental Investigation of
Nonaqueous Phase Liquid Dissolution in Saturated Subsurface Systems: Steady
State Mass Transfer Rates”, Water Resourc. Res., 28(10): 2691-2705.

Priddle, M.W., and K.T.B. MacQuarrie. 1994. "Dissolution of Creosote in Groundwater:
An Experimental and Modeling Investigation”, J. Contam. Hydrol, 15:27-56.

174

Reichmuth, DR 1984. ”Subsurface Gasoline Migration Perpendicular to Ground Water
Gradient - A Case Study”, Proceedings of the NWWA/API Conference on Petroleum
Hycbocarbons and Organic Chemicals in Ground Water - Prevention, Detection
and Restoration, Houston, Texas, pp. 43-52.

Rosen, MJ. 1989. Surfactants and Interfacial Phenomenon, 2nd Ed, John Wiley and
Sons.

Roy F. Weston Inc. 1988. Remedial Technologies for Leaking Underground Storage
Tanks, Lewis Publishers, pp. 55-63.

Sale, T.C., Piontek, KR, and M.Pitts. 1989. ”Chemically Enhanced In Situ Soil
Washing”, Proceedings of the Petroleum Hydrocarbons and Organic Chemicals in
Groundwater:Prevention, Detection and Restoration, pp. 487-503.

Schlitt, W.J., and DA Shock. 1979. ”In Situ Uranium Mining and Groundwater
Restoration”, Proc. of the New Orleans Symposium, Solution Mining Committee,
Society of Mining Engineers of American Institute of Mining, Metallurgical and
Petroleum Engineers.

Schmelling, S.G. 1992. "An Overview of Aquifer Remediation: NAPLS, Pump and Treat
and Bioremediation", Subsurface Contamination by Immiscible Fluids, K.U. Weyer,
Ed, AA Balkema/Rotterdam/Brookfield, pp. 39-45.

Schwille, F. 1988. Dense Chlorinate Solvents in Porous and Fractured Media: Model
Experiments, Translated from German by J.F . Pankow, Lewis Publishers, Chealsea,
Michigan.

Sherwood, T.K., Pigford, R.L., and CR Wilke. 1975. Mass Transfer, McGraw-Hill.

Soeryantono, H., Wallace, RB., and TC. Voice. 1994. ”Acceptability of Pressure
Equilibrium Method”, Draft, Michigan State University, Department of Civil and
Environmental Engineering.

Souza, WR 1987. Documentation of a Graphical Display Program for the Saturated-
Unsaturated Transport (SUTRA) F inite-Element Simulation Model, U.S.G. S.
Water-Resources Investigations Report 87-4245.

van der Heijde, P.K.M., El-Kadi, AL, and SA Williams. 1988. ”Groundwater Modeling:
An Overview and Status Report", EPA/600/2-89/028.

van Genuchten, MT. 1980. ”A Closed Form Equation for Predicting the Hydraulic
Conductivity of Unsaturated Soils”, Soil Sci. Soc. Am. J., 44:892-898.

van Genuchten, MT. and W.J. Alves. 1982. Analytical Solutions of the One-
Dimensional Convective-Dispersion Solute T ransprot Equation, U.S.G.S. Technical
Bulletin No. 1661, 149 p.

175

Verschueren, K. 1983. Handbook of Environmental Data on Organic Chemicals, 2nd
Ed, Van Nostrand Reinhold, pp. 595-596.

Voss, CL 1984. A F mite-Element Simulation Model for Saturate- Unsaturated F luid-
Density-Dependent Groundwater Flow with Energy Transport or Chemically-
Reactive Single-Species Solute Transport, USGS Water-Resources Investigation
Report 84-43 69.

Wang, HF. and MP. Anderson. 1982. Introduction to Groundwater Modeling: Finite
Drflerence and Finite Element Methods, W.H. Freeman.

Wardlaw, NC. 1982. "The Effect of Geometry, Wettability, Viscosity, and Interfacial
Tension on Trapping in Single Pore-Throat Pairs”, J. Can. Petrol. Tech, 21(3):21-
27.

Weber, W.J.,Jr., McGinley, PM, and L.M. Katz. 1991. ”Sorption Phenomena in
Subsurface Systems: Concepts, Models and Effects of Contaminant Fate and
Transport", Water Res., 25(5):499-428.

Wheatcraft, SW. and F. Winterberg. 1985. "Steady State Flow Passing Through a
Cylinder of Permeability Different From the Surrounding Medium", Water Resourc.
Res., 21(12):]923-1929.

Wilson, J .L. and SH. Conrad. 1984. ”Is The Physical Displacement of Residual
Hydrocarbons a Realistic Possibility in Aquifer Restoration”, Proceedings of
Conference on Petroleum Hydrocarbons and Organic Chemicals in Ground Water,
NWWA, Houston, Tex., , Nov. 1984, pp. 274-298.

“Wilson, J.L., Conrad, S.H., Hason, WR, Peplinski, W., and E. Hagen,. 1990.
”Laboratory Investigation of Residual Liquid Organics fi'om Spills, Leaks, and the
Disposal of Hazardous Wastes in Groundwater”, EPA/600/6-90/004.

Mlson, DJ. 1989. "Soil Clean Up by In-Situ Surfactant Flushing. 1. Mathematical
Modeling”, Sep. Sci. T echnol., 24(11): 863-892.

Wilson, J.L. 1994. "Visualization of Flow and Transport at the Pore Level”, Transport
and Reactive Processes in Aquifers, Th. Dracos and F. Staufl‘er, Eds, Rotterdam;
Balkema.

“Witco. 1991. Facsimile received from WITCO MPO ORG SLS, Nov. 26, 1991.

Zalidis, G.C., Annable, M.D., Wallace, RB., Hayden, N.J., and TC. Voice. 1991. "A
Laboratory Method for Studying the Aqueous Phase Transport of Dissolved
Constituents from Residually Held NAPL in Unsaturated Soil Columns”, J. Contam.
Hydrol, 8:143-156.

   

”’IlllilllIIIIIII‘“