Download as docx, pdf, or txt
Download as docx, pdf, or txt
You are on page 1of 13

Final Report on Simulation of groundwater and chloride concentrations

at the Borden landfill

The objectives of the report is:

1. Simulate 2-D profile steady state groundwater flow simulation of the site
using the finite difference scheme implemented in excel.
2. Solve for the 3-D flow field using MODFLOW.
3. Solve for the chloride profile in 1979 using particle tracking implemented in
excel in for your 2D groundwater simulation. Use a random fluctuation in
ground water velocity to model dispersion.
4. Estimate the chloride profile in 1979 using an analytical equation.
5. Estimate the chloride profiles using the MODFLOW contaminant transport package.

Executive summary:

According to the uploaded Frind and Hokkanen paper and the modeling strategy employed here
, this report simulates head dispersion and chloride distribution beneath the Borden landfill:

1. For the 2D finite difference model for groundwater flow in excel, the values
considered for different parameters is listed in the table below.

Recharge 15 cm/yr
Hydraulic Conductivity 1.17 X 10-4 m/s
Thickness 22 m
Δx 50 m
Δz 2m

And the 2D finite difference equation that is used here for the solution is one
with recharge and other without recharge:
2. The following initial specifications for MODFLOW modeling were taken into c
onsideration:

active transportation Model for retardation: linear isotherm, MT3DMS is the trans
port engine. No kinetic response occurred. USGS MODFLOW 2005 from WH is the f
low engine. Initial head = 220 m, Specific yield = 0.2, Contaminant Initial Concent
ration = 400 mg/l, Kd = 0 l/mg, Start Date: January 1, 1940. Kx=ky=kz=0.000117
m/s (Considering Homogeneous Medium). The head is determined in the cell's
center for each individual cell (node).

Defining layer elevations:

Our model has two layers: an unconfined aquifer on top and an aquitard on the bottom.
We have determined that the initial head of an unconfined aquifer is 220 meters (1st layer), an
d that the initial head for an aquitard is 203 meters (2nd layer).
The transient groundwater flow model was developed using MODFLOW-2005.
The finite difference model has 12 rows, 22 columns, and 2 layers (Unconfined aquifer and aqu
itard).
There are 20 X 20 cells in the total grid.
Under these conditions, the grid cells have a constant size of 50 meters, occupying an area of
about 106 meters.
Since chloride doesn't sorb or degrade and hence functions as a conservative contaminant in th
e model, it is believed to be the

Layer 1 conductivity: 0.000117 m/s (Aquifer which is considered homogeneous here carrying
steady state flow)
Layer 2 conductivity: 10^-10 m/s (Aquitard, as we know it has lower permeability).

Additionally, this information is used in modeling to determine how the plume disperses across
time and space in relation to groundwater movement.
All other boundaries are no flow since no flow is the default boundary condition.
So there is no need to add a head on the right side.
In the section titled "Boundary conditions," the boundary conditions are applied in accordance
with the research paper's presentation.
Dispersion is 6 m, Ponding is 0.3 m, and Recharge is 150 mm/year added.
on layer 1 (the unconfined aquifer) at the top;
Constant head: 220 meters from beginning to end.
Therefore,
Setup for MODFLOW-2005:
For the best simulation, a run type of steady-state with a steady state simulation time of 14,60
0 has been taken into account.
The simulation time length is a standard MT3DMS configuration definition.
Changed Words
Structural Changes
Longest Unchanged Words
3. The data used in particle tracking implemented in excel in for 2D groundwater
simulation is as follows:

R 4.75E-09 m/s
Ksat 0.000117 m/s
b 22 m/s
Δx 50
Δ x2 2500
Δz 2
Δ z2 4
2(Δ x2 + Δ z2) 5008
R *Δ z *Δ x2/K 0.202991453

X no. of blocks 22
Z no. of blocks 12

Particle tracking was used to replicate the chloride concentration over 40 years. For particle
tracking, a total of 100 particles were taken into account starting in the beginning of the era,
in 1940, and simulated until 1979. Every particle has a specific seepage velocity
that was determined by the head difference computed using a 2D finite difference model.
Each particle was given a random velocity fluctuation magnitude for dispersion. Each particle's
travel distance over a specific period of time was monitored, and a graph was created.

10203040
6
0

5
0

4
0
PARTIC

3
0
0 10 20 30 40 500 60 70 80 90 100
2
0 0 0 0 DISTAN 0 0 0 0 0
0
CE

1
0

4. For estimating chloride profile in 1979. The data used for solving the problem
and the analytical equation that is used is:
Based on the analytical solution performed the graph was plotted between distance and
concentration. Similarly, concentration vs time. Compared and nalysed with the observation in
the paper given, it

5. The conditions and considerations with outcomes are presented in 2nd answer
above. The results and discussion for both 2nd and 5th individually will be
presented in the conclusion and discussion further.
Background

Why Borden Site?

From 1940 through 1976, the Borden landfill was in use on a regular basis. The dumping of
trash began over a small sand hill and grew slowly in a southerly direction. Up until 1973, the
site was used as an open landfill; following that, sophisticated landfill procedures were used
until closure. The finished landfill has a surface area of 5.4 acres and a thickness of 5 to 10 m.
The Canadian Forces Base Borden, a sizable military installation located 60 kilometres
northwest of Toronto, Canada, was the location of the Borden Experimental Site for Studies of
Groundwater Contamination until it was founded in 1978 (Cherry et al., 1996). The
experimental site is largely utilised for controlled studies in which a range of industrially
produced chemicals, some of which are dangerous compounds, are carefully released into the
subterranean environment while their movement and fate are meticulously tracked. The
Borden site has contributed significantly to the improvement of information regarding the
behaviour, destiny, and cleanup of several hazardous substances that are frequently found in
groundwater in industrial regions of North America and Europe for more than ten years
(Anderson et al., 1979). The site is anticipated to be used heavily throughout the remaining
period of the decade and maybe longer.
This site qualifies as a test case for the assessment of numerical transport models due to its
properties. The location is hydrogeologically straightforward, the polluted plume has been
closely observed, the aquifer's features have been determined, and the aquifer's dispersive
properties have been investigated using data from the article, independent of the plume. The
actual plume spreads both horizontally and vertically.

Introduction

The employment of models to predict the movement of pollutants in groundwater systems has
received more attention in recent years due to the expanding popularity of subterranean
waste disposal. The effects of dispersion have been incorporated into contamination transport
models that have been used in a variety of real-world settings. Advection, dispersion (including
diffusion), and chemical reaction make up the transport process. The contamination plume is
located in an abandoned landfill on the Canadian Forces Base in Borden, Ontario, in a shallow
aquifer (Zhao et al., 2005). The landfill is situated above an unconstrained glaciofluvial sand
aquifer. Under the landfill, the aquifer is roughly 20 m thick and thins out to around 9.5 m
towards the north. The vast deposit of sandy and clayey silt that the unconfined aquifer sits
upon serves as a physical barrier to the flow of pollutants. Hydraulic conductivity values were
calculated using permeameter measurements on Borden sand cores, with a mean value that
ranged from 1.17 x 10 -2 cm/s to 8.7 x 10 -3 cm/s. The Borden sand's porosity was estimated
to be 0.38. The majority of the flow is said to be toward the north during the whole year.
Increased spring recharge builds up the water table beneath the landfill, increasing both the
downward and outward flow away from the dump. The mound remains missing the remainder
of the year and disappears in the summer. The natural vegetation and unaltered regions
outside the landfill are thought to recharge on average at a rate of between 10 and 20 cm per
year. According to observations made in 1979, the plume travels northward for roughly 750
metres from the dump. In addition, the plume's core contains high- concentration peaks that
may be the consequence of changes in the source's power over time. Although it occupies the
middle to lower part of the aquifer, it almost fills the whole thickness of the aquifer beneath
the landfill before getting thinner as it descends. A weak vertical dispersion mechanism is
indicated by the plume's sharp vertical gradient, which shows concentrations changing quickly
across close vertical distances at the top and bottom surfaces. In addition, the plume's core
contains high-concentration peaks that may be the consequence of changes in the source's
power over time. Recharge has a high value of 55 cm at the landfill, a low value of 10 cm at
the northern end, and a base value of 15 cm in the centre and southern areas.

Error types occur with any numerical solution of the transport equation

The well-known grid Courant and Peclet criteria are applied to each coordinate direction to
limit the first of these mistakes, numerical dispersion. The Peclet requirements are
Even after the Peclet and Courant conditions have been fully met, the second form of error
that caused by the flow lines' divergence from the grid lines can still happen. Keeping the
angle as minimal as feasible can help to minimise this mistake, which is at its lowest point
when the flow lines and grid lines meet at a 45° angle. The grid must, however, be perfectly
aligned with the flow direction throughout in order to achieve complete elimination; this is
referred to as a principle direction scheme. This kind of mistake is anticipated to be present in
the simulation results inevitably since the flow direction in the Borden cross-section
continuously changes from virtually vertical to horizontal. Similar effects will also be seen by
other number methods.
The third class of error is brought on by the coordinate transformation's Jacobian's tiny nonzero
terms.
Due to the tiny size of the solution matrices, the fourth class of error, known as machine,
round off, is identical to that of a traditional alternating direction finite difference scheme but
is less significant than that of a conventional finite element scheme.
At the Borden site, longitudinal dispersivities range from around 5 to 10 m. The huge
concentration gradients in the vertical direction imply that transverse vertical dispersivity will
be substantially less. We employed a longitudinal dispersivity range of 6 m for the parametric
analysis.
The 5 x 10 -5 m2/day value of the molecular diffusion coefficient D* is maintained.

Boundary Conditions

The water table serves as the model's top boundary, and the simulation is only allowed to
cover the saturated zone. The concentration at the border of the flow system is therefore
assumed to always be equal to that of the source, and this allows the source to be represented
by a first-type boundary condition (defined head). Another option is to utilise a third sort of
boundary condition,
which determines the mass flow of contaminants entering the system but leaves the border
concentrations unconstrained.

Constant Head
No flow BC

No flow BC

The third-type boundary continuously adds mass to the system, whereas the first-type
boundary fixes the concentration at the boundary. Under either boundary condition, the plume
does not cross the water table boundary off-source due to its low transverse dispersivity of
0.01 m.

Modelling Approach

Using MODFLOW-2005, the transient groundwater flow model was created. 12 rows, 22
columns, and 2 layers make up the finite difference model (Unconfined aquifer and aquitard).
The entire grid is composed of 20 X 20 cells. The grid cells have a consistent size of 50 metres
under these circumstances, covering an area of around 106 metres. The cell size was
determined by taking into account three different aspects: (1) computing efficiency, (2)
accurate representation of the data, and (3) successful modelling of groundwater flow on a
regional scale (Zhou et al., 2011). From where layer 2 really starts, layer 1's thickness rises
from 220 masl at the top to 198 masl at the bottom, with a total depth of 22 m (average value
taken into account). For examining groundwater flow patterns and evaluating groundwater
resources, the model was calibrated under a steady state. Additionally, in order to examine
groundwater flow systems, a 3D transient groundwater flow mode1 was created for the Borden
landfill. The simulation period for the transient model lasted from 1990 to 1976.
Application of 3D groundwater flow models has significantly advanced regional flow system
analysis. Due to its adaptable modular design, thorough coverage of hydrogeological processes,
and free public domain availability, MODFLOW has since its 1988 release become the industrial
benchmark for groundwater modelling around the world. MODFLOW-88 had incremental
updates to become MODFLOW-2000 (Harbaugh et al., 2000) and MODFLOW-2005 (Harbaugh et
al., 2005). The simulation of saturated-unsaturated flow processes, groundwater simulation-
optimization processes, irrigation processes, density dependent flow processes, parameter
optimization processes, and solute transport processes are all included in the MODFLOW
family's 2005 iteration.
The fundamental process is detailed in depth below because the majority of the models that
will be examined have been used at operational sites. In this example, there are two phases
involved in using an advection-dispersion model: the calibration phase, which comprises the
following steps:

1. To calibrate the groundwater flow model, hydraulic conductivity values and sometimes flow
rates at boundaries, sources, and sinks are changed until the model accurately resembles the
observed heads.

2. Darcy's Law is used to calculate groundwater velocities.

3. To calibrate the contaminant transport model, dispersivities and occasionally porosity are
changed until the model simulates the reported concentrations.

Conclusion

The most crucial step in accurately calibrating the models used to the reported groundwater fl
ow and transport conditions is to simulate Borden landfill sites using analytical models.
Sensitivity tests are carried out on linked two- and three-dimensional systems using the MODFL
OW 2005 model.
The groundwater flow and contaminant transport conditions at the dump site were effectively
simulated using this model, which has shown to be a useful tool.
However, when compared to the supplied paper, the numbers from our analytical solution reve
al a significant difference in the outcome: the graph and illustrations in the analytical solution
are more numerous than they are in the research article. The diffusion has been seen to rise as
time has gone on, and the same is true of velocity.
The disadvantage in our analysis was that, as the size increases, the diffusion will likely grow t
he most and might potentially lead to more system-wide plume spreading.
Using MODFLOW, the groundwater and pollutant flow has been designed and is displayed above
. Since the plume is likely to be different after a 40year period if we consider a continuous sour
ce of contamination. Additionally, employing an analytical method to trace particles would be
extremely challenging. Advection
dispersion and advection models are anticipated to remain helpful in the future when analyzing
groundwater contamination issues.

References
1. Anderson, Mary P.; Cherry, John A. (1979). Using models to simulate the movement of
contaminants through groundwater flow systems. C R C Critical Reviews in
Environmental Control, 9(2), 97–156. doi:10.1080/10643387909381669.

2. Cherry, J. A., Barker, J. F., Feenstra, S., Gillham, R. W., Mackay, D. M., & Smyth, D. J.
A. (1996). The Borden Site for Groundwater Contamination Experiments: 1978–1995.
Groundwater and Subsurface Remediation, 101–127. doi:10.1007/978-3-642-45750-0_7.

3. A.W. Harbaugh. MODFLOW-2005, the U.S. Geological Survey’s Modular Ground


Water Flow Model – The Groundwater Flow Process. Techniques and Methods 6–A16,
U.S. Geological Survey, Reston, Virginia (2005).

4. A.W. Harbaugh, E.R. Banta, M.C. Hill, M.G. McDonald. The U.S. Geological Survey’s
Modular Ground Water Flow Model – User Guide to Modularization Concepts and the
Ground Water Flow Process. U.S. Geological Survey (2000).

5. Xianda Zhao, Roger B. Wallace, David W. Hyndman, Michael J. Dybas, Thomas C.


Voice. (2005). Heterogeneity of chlorinated hydrocarbon sorption properties in a sandy
aquifer, Journal of Contaminant Hydrology, Volume 78, Issue 4, 2005, Pages 327-342,
https://doi.org/10.1016/j.jconhyd.2005.06.002.

6. Yangxiao Zhou, Wenpeng Li. A review of regional groundwater flow modeling,


Geoscience Frontiers, Volume 2, Issue 2, 2011, Pages 205-214.
https://doi.org/10.1016/j.gsf.2011.03.003.

You might also like