Vous êtes sur la page 1sur 26

Groundwater

Modelling Task
January 2015

[Escriba aqu una descripcin breve del documento. Normalmente, una


descripcin breve es un resumen corto del contenido del documento. Escriba aqu
una descripcin breve del documento. Normalmente, una descripcin breve es un
resumen corto del contenido del documento.]

Table of Contents
1.

INTRODUCTION................................................................................................. 2

2.

CASE STUDY AND PROBLEM STATEMENT..........................................................2


Description of Study Area and Available Data......................................................2
Problem Statement.............................................................................................. 3
Objectives............................................................................................................ 3

3.

4.

5.

METHODOLOGY................................................................................................. 4
3.1.

General....................................................................................................... 4

3.2.

Tools............................................................................................................ 5

GROUNDWATER FLOW MODEL..........................................................................6


4.1.

STEADY STATE SIMULATION........................................................................6

4.2.

TRANSIENT SIMULATION.............................................................................8

4.3.

CALIBRATION OF TRANSIENT MODEL........................................................10

TRANSPORT MODEL......................................................................................... 14
5.1.

TRANSPORT BY ADVECTION......................................................................14

5.2.

TRANSPORT BY ADVECTION AND DISPERSION..........................................14

5.3.

EFFECT OF EFFECTIVE POROSITY AND DISPERSION COEFFICIENT............17

6.

CONCLUSIONS................................................................................................. 19

7.

REFERENCES................................................................................................... 20

8.

ANNEX A.......................................................................................................... 20

List of Tables
Table
Table
Table
Table
Table
Table

1. Observed hydraulic heads after 100 days of pumping..................................3


2. Summary of Layer Properties for Steady State simulation............................7
3. Layer Properties for Transient Simulation......................................................9
4 Summary of input parameters.......................................................................9
5. Summary of Manual Calibration Trials.........................................................11
6 Summary of Statistics for calibration...........................................................13

List of Figures
Figure 1. Profile of Aquifer.......................................................................................... 3
Figure 2. Layer 1: Layout and Observation wells........................................................4
Figure 3. Layer 3: Layout and Observation wells........................................................4
Figure 4. Methodology to build Groundwater Flow model and Solute Transport model
................................................................................................................................... 5
Figure 5. Soil type and boundaries in Layer 1 (left), Layer 2 (center) and Layer3
(right)......................................................................................................................... 8
Figure 6. Hydraulic Head simulated in Steady State conditions for Layer 1 (Left),
Layer 2 (Center) and Layer 3 (Right)..........................................................................9
Figure 7. Cross Section profile of the first 100 m of the West boundary.....................9
Figure 8. Hydraulic Heads after 100 days of Pumping, Layer 1 (Left), Layer 2
(Center) and Layer 3 (Right)..................................................................................... 11
Figure 9. Scatter plot of Observed Vs. Simulated Hydraulic Heads before calibration
................................................................................................................................. 11
Figure 10. Hydraulic Head contour after Manual Calibration, Layer 1 (Left), Layer 2
(Center) and Layer 3 (Right)..................................................................................... 12
Figure 11. Scatter Plot of Observed Vs Simulated Hydraulic Heads after Manual
Calibration................................................................................................................ 12
Figure 12. Scatter Plot of Measured and Simulated Hydraulic Heads.......................13
Figure 13 . Hydraulic Heads for Summary calibration...............................................14
Figure 14 Particle Tracking for Transient flow without Pumping well (right) and with
Pumping well (left).................................................................................................... 15
Figure 15 Pollutant concentration for Layer 1 (left), Layer 3 (right) for the Case
without Pumping Well............................................................................................... 16
Figure 16 Pollutant concentrations observed in Well after 20 years without pumping
................................................................................................................................. 17
Figure 17. Pollutant concentration for Layer 1 (left) and Layer 3 (right) for the Case
with Pumping Well.................................................................................................... 17
Figure 18 Pollutant concentrations observed in Well after 20 years with pumping...18

Figure 19 Pollutant concentration for Layer 1 (with pumping well) for D L=10 [m]
(left) and 50 [m] (right) after 20 years.....................................................................19
Figure 20 Pollutant concentration and Concentration-Time curve at V1 for Layer 1
(with pumping well)
for n e=0.25 (sand) and 0.01
(clay) (up) and ne=0.5 (sand) and 0.05 (clay) (down)..............................................20

GROUNDWATER MODELLING TASK


1. INTRODUCTION
Groundwater is exploited for water drinking supply, irrigation and industrial
purposes; the increasing demand due to population growth has led to the
development of management and protection technologies. In that context,
groundwater models that can properly represent quantity and quality of the
resource are widely used.
Pollution of groundwater can be caused by sub-surface waste disposal, seepage of
contaminants from rivers and lakes, impoundment of toxic waste on unlined
surfaces, indiscriminate spraying of insecticides, pesticides and excessive use of
chemical fertilizers, etc.(Thangarajan M., 2007) The polluted groundwater flows
through the aquifers and depending on the hydrogeological properties the time
travel can range from days to years. The exploitation of unidentified polluted
groundwater is a huge threat for users.

2. CASE STUDY AND PROBLEM STATEMENT


Description of Study Area and Available Data
The study area has an extent of 2000 m by 2000 m and encloses an aquifer
composed by three layers: two sand layers and a clay layer in between. According
to observations from A3, B3 and W1 (observation wells installed in Layers 1 and 3
depicted in Figure 2 and Figure 3) there is also a small part of sand in Layer 2 as
shown in Figure 1.
The spatial boundaries of the study area are bedrock at North and South. The West
boundary consists of a 100 m wide ditch which results in continuous and constant
water input to the aquifer corresponding to a recharge rate of 0.02 m/d in this area.
The entire area has an average recharge of 0.0001 m/day. The East boundary
consists of a 100 m wide river. The river water level is 35 m (above the bottom of
the aquifer) in the North and decreases to 30 m in the South. The river has an
average depth of 2 m and a river bed consisting of 2 m sand and fine gravel with a
vertical hydraulic conductivity of 30 m/day.
Hydraulic conductivity of sand is estimated between 20 and 50 m/day. For clay,
hydraulic conductivity is considered to be lower than 5 m/day. The storage
coefficient of sand and clay is estimated as 0.1.

A pumping well located at coordinate (1450 m, 550 m) pumps at 2000 m/day. After
100 days of pumping, hydraulic heads are measured in the monitoring wells, the
values are indicated in Table 1.

Figure 1. Profile of Aquifer

Figure 2. Layer 1: Layout and Observation


wells

Well
name
W1
X1
Y1
Z1
V1

Layer 1
X
Y
Observed head
(m)
100 100
35.2
0
0
130 130
34.7
0
0
145 750
33.8
0
170 250
32.1
0
155 450
33
0

Figure 3. Layer 3: Layout and Observation


wells

Well
name
A3
B3
C3
D3
E3

Layer 3
X
Y
Observed head
(m)
750 850
35.6
115
0
150
175
0
125
0

105
0
105
0
550

34.9

550

34.2

37.2
32.3

Table 1. Observed hydraulic heads after 100 days of pumping

Problem Statement
At the north-west of the study area (Figure 2) an industrial zone is source of
contamination and causes groundwater pollution in the aquifer. It is required to
determine if this groundwater pollution threatens the groundwater quality in the
river or in drinking water extraction well in layer 3.
Objectives
The main purpose of this work is to build a calibrated groundwater flow model and
solute transport model. The specific objectives are:
Groundwater Flow Model:

Describe and substantiate the different modelling steps and the conceptual
model of this hydrogeological model.
Build a transient model.
Calibrate the transient model using the measured heads from Table 1. Use
statistical indicators, hydraulic head contour maps and hydraulic head scatter
plots to judge calibration.

Solute Transport Model

Use particle tracking to determine the transport time and direction of polluted
groundwater by advection with and without pumping in the pumping well.
Determine if polluted groundwater is flowing to the river or to the pumping
well and how long does it take for the pollution to reach the river or the
pumping well.
Calculate the concentration distribution of pollutants by transport by
advection and dispersion with and without pumping in the pumping well and
compare with particle tracking results.
Calculate and explain the effect of effective porosity and dispersion
coefficient on groundwater pollution.

3. METHODOLOGY
3.1.

General

The general methodology followed in this work is shown in Figure 4.


GROUNDWATER FLOW MODEL
Inpu
t

Stead
y
State

Initial
h

Transien
t

Manual
Calibrati
on
Trial
and
Error

Automati
c
Calibrati
onPES
T

Statistics
h contours
maps and
Scatter plots

Best
Parameter
s

SOLUTE TRANSPORT MODEL


-Pumping
-No
pumping

By Advection

-Flow
Direction

-Pumping
-No
pumping

By Advection
and
Dispersion

-Flow
Direction
-Time
-Flow
Direction
-Time
-

Changes in:
-Effective porosity
-Dispersion
coefficient

By Advection
and
Dispersion

Figure 4. Methodology to build Groundwater Flow model and Solute Transport model

3.2.

Tools

MODFLOW: Modflow is a modular three-dimensional finite-difference groundwater


model published by the U. S. Geological Survey. The code is divided into entities
called processes and each process is related to a specific equation. The calculation
of hydraulic heads is carried by solving one finite differences equation for each cell,
expressing the relationship between the head at a node and the heads at each of
the six adjacent nodes at the end of a time step. (Simcore Software, 2012)
One of the packages that PM supports for solving systems of simultaneous linear
equations is the Preconditioned Conjugate-Gradient Package 2 PCG2. For this
package the preconditioning method of Modified incomplete Cholesky was specified.
The convergence criteria is based on Head Change and Residual, when the
maximum absolute value of head change at all nodes during an iteration is less
than or equal to the specified Head Change and the maximum absolute value of the
residual at all cells is less than or equal to Residual, iteration stops. (Simcore
Software, 2012)
PEST: Pest is a tool used that allows performing parameter automatic calibration,
by adjusting model parameters in order to minimize the differences between modelgenerated values and measurements. This is accomplished by the definition of an
Objective Function (Sum of Squares Error for this case) that has to be minimized.
PMPATH: PMPATH is an advective transport model that uses the semi-analytical
particle-tracking scheme presented by Pollock to calculate the groundwater paths
and travel times. It assumes that each velocity component varies linearly within a
model cell in its own coordinate direction. The model can be used to simulate

advective transport in groundwater, to delineate contaminant capture zones,


injection zones and wellhead protection areas or to find the point of origin of water
in specified zones. (Simcore Software, 2012)
MT3DMS/SEAWAT: The MT3DMS/SEAWAT is a transport package that simulates
advection and diffusion processes. In this case, the advection scheme is solved with
the third-order TVD method based on the Ultimate algorithm. (Simcore Software,
2012) The algorithm consists of an explicit conservative control-volume formulation
that makes use of a universal limiter for transient interpolation; the advantage of
the algorithm is that it can be used with explicit schemes of arbitrarily high order
suppressing numerical oscillations without corrupting the resolution.(B.P. Leonard,
1988)
According to Modflow Manual the concentration change due to dispersion is solved
with a fully explicit central finite-difference scheme. The stability criterion is based
on the transport step size, t is calculated for each active cell and the minimum t
is taken as the maximum allowed step size for solving the dispersion term. The
upper limit of t is equal to:
Equation 1

Where x, y and z correspond to the size of the grid, R is the retardation factor
and Dxx, Dyy, and Dzz are the components of the hydrodynamic dispersion
coefficient that can be calculated with:

Equation 2

Where L [L] is the longitudinal dispersivity; TH [L] is the horizontal transverse


dispersivity; TV [L] is the vertical transverse dispersivity; vx, vy, and vz [LT-1] are
components of the flow velocity vector along the x, y, and z axes. Longitudinal
Dispersivity is used to approximate the spreading of the solute concentration in
groundwater caused by the irregular shape of the interconnected pore space and
the velocity variations at the microscopic level as well as the unresolved
macroscopic level.
(Simcore Software, 2012)

4. GROUNDWATER FLOW MODEL


4.1.

STEADY STATE SIMULATION

Conceptual Model
The Steady State model was run to obtain Initial Prescribed Conditions for Transient
Model. The following conceptual model choices were considered:

Steady-state model
Model dimensions of 2000 m x 2000 m that correspond to natural boundaries
and Total thickness of 40 m. The grid size was set to 100 m x 100 m in order
to obtain a reasonable degree of accuracy and also considering the
dimensions of the ditch at the East and river at the West.
Layer characteristics (Table 2): The Second and Third Layers were defined as
Confined/Unconfined because of the sand hole in the Second Layer which
fully connects them with the upper and lower layer.

Property
First
Second
Third

Unconfined
Confined/Uncon
fined
Confined/Uncon
fined

Top
Elevation
40

Bottom
Elevation
20

20

10

10

Table 2. Summary of Layer Properties for Steady State simulation

Parameters

In respect to Time parameters, Steady State mode was specified.


Figure 4 depicts the soil type for every Layer, assumed values of Hydraulic
Conductivity for sand equal to 35 [m/h] and for clay equal to 3.5 [m/h] were
assigned, these average values will be calibrated afterwards. In respect to Vertical
Hydraulic Conductivity, for both sand and clay 10% of Horizontal Conductivity was
assumed.
The effective porosity was assumed equal to ne=0.25 for sand according to typical
values (Stephens et al., 1998) and ne=0. 01 for clay (Johnson, 1967).
Taking into account the lowest water level in the River boundary, the Initial and
Prescribed conditions were defined equal to 30 m over the entire aquifer.

Boundary Conditions

As depicted in Figure 5, the Boundary conditions were: active cells in every cell of
the model, the no-flow boundaries at the North and South of the aquifer are defined
by the limit of the Area.

San
d

Cla
y

Figure 5. Soil type and boundaries in Layer 1 (left), Layer 2 (center) and
Layer3 (right)

River: At the East boundary the effect of flow between the river and the aquifer
were simulated with the River Package, the input required was Head in the river [m]
interpolated between 35 m at N and 30 m at S, Elevation of the Riverbed bottom
[m] equal to Head minus river depth and river bed (4 [m]) and River hydraulic
conductance CRIV [m2/h] calculated with the following formula:

Criv =

K rivLW riv
M riv

Equation 3

Where Kriv is the Hydraulic Conductivity of Riverbed [m/h] and is equal to 30


[m/day], L is the length of the river within a cell [m] equal to 100 [m], Wriv is the
Width of the river [m] equal to 100 [m] and Mriv is the Thickness of the riverbed [m]
equal to 2 [m]. After replacing these values we obtain a River hydraulic conductance
equal to 150000 [m2/h].
Recharge: There is constant recharge at the entire top of layer 1 equal to 0.0001
[m/day] except in the ditch along the West side where it is equal to 0.02 [m/day].
Results

The model was run in Steady State mode; the resulting hydraulic heads for each
layer are presented in Figure 6. The values range from 30m at the South limit of the
river and 37m at the Northwest corner of the study area. In Figure 7, the presence
of clay in Layer 2 is represented by the change of direction of equipotential lines
from vertical to almost horizontal. Due to the sand hole located in the middle of
Layer 2, the bending of equipotential lines is only clear in the furthest sides of the
model.

Figure 6. Hydraulic Head simulated in Steady State conditions for Layer 1 (Left), Layer 2
(Center) and Layer 3 (Right)

Figure 7. Cross

Section profile of the


first 100 m of the West boundary

4.2.

TRANSIENT SIMULATION

Conceptual Model
The following conceptual model choices were considered:
Transient-state model
Model dimensions are the same as for Steady State. In respect to the grid
size, where pumping well and observation boreholes are located, it was
reduced to 50 m x 50 m to increase the accuracy of hydraulic head
calculation in those regions.
Layer characteristics: The geometrical properties of the layers are the same
as for Steady State model. During a transient simulation the rate of change in
storage must be specified, thus, as indicated in Table 3 the input mode for
Storage Coefficient was switched to User Specified for Layers 2 and 3.

Layer

Property

First
Secon
d

Unconfined
Confined/Unconfined T
varies
Confined/Unconfined T
varies

Third

Storage
Coefficient
Calculated
User Specified
User Specified

Table 3. Layer Properties for Transient Simulation

Parameters

In respect to Time parameters, a transient period with duration of 100 days and a
time step of 2 days was specified.
Horizontal and Vertical Hydraulic conductivity for sand and clay were the
parameters to be calibrated. Effective porosity values were not changed from the
typical values specified for Steady state model.
The Initial and prescribed hydraulic Head values were loaded from the results of the
Steady State model.
The storage term for a confined layer is given by storativity, it depends on the
compressibility of the water and the elastic property of the soil matrix; it is
calculated with Equation 4:

S=Ssb

Equation 4

Where b is the Layer thickness [m] and Ss is the specific storage [m -1] defined as
the fraction of volume of water that a unit column of aquifer releases from storage
under a unit decline in hydraulic head.
In a phreatic or unconfined aquifer, the storage term is given by specific yield or
drainable porosity. Specific yield is defined as the volume of water that an
unconfined aquifer releases from storage per unit surface area of aquifer per unit
decline in the water table; it is a function of porosity. (Simcore Software, 2012)
In our model, Specific Yield was used to calculate rate of change in storage for Layer
1. For Layers 2 and 3 which are convertible between Confined and Unconfined with
Constant transmissivity, Specific storage equal to 0.1 was used when the layer is
fully saturated and specific yield when it is not.
In Table 4 a summary of all input parameters is presented.

Parameters
Horizontal Hydraulic Conductivity Sand
Horizontal Hydraulic Conductivity Clay
Vertical Hydraulic Conductivity Sand

Units
[m/day]
[m/day]
[m/day]

Initial
35
3
3.5

Vertical Hydraulic Conductivity Clay


Effective porosity Sand
Effective porosity Clay
Storage Coefficient Sand
Storage Coefficient Clay
Specific Yield Sand
Specific Yield Clay

[m/day]

0.3
0.25
0.01
0.0001
0.02
0.25
0.01

Table 4 Summary of input parameters

Boundary Conditions

Boundary condition, recharge and river characteristics are the same as for Steady
state model. A pumping well is installed in layer 3 in the position (1450 m, 550 m)
with injection equal to negative 2000 [m3/day] and the 10 observation boreholes
(Table 1) are drawn in the aquifer.
Results
The model was run in Transient Flow conditions with pumping during 100 days,
hydraulic heads at the end of the period are shown in Figure 8. The pumping well
produces a drawdown of 2 m. An initial comparison between observed and
calculated hydraulic heads is presented in the Scatter plot in Figure 9, the degree of
dispersion from the bisector and the low value of the variance suggests that the
assumed initial parameters need small adjustment. In general, hydraulic head is
underestimated and more specifically in wells V1, Y1, E3 (near the well) and C3 in
less than 0.5 m, this underestimation indicates that Hydraulic Conductivity should
be increased.

Figure 8. Hydraulic Heads after 100 days of Pumping, Layer 1 (Left), Layer 2 (Center) and
Layer 3 (Right)

Figure 9. Scatter plot of Observed Vs. Simulated Hydraulic Heads before calibration

4.3.

CALIBRATION OF TRANSIENT MODEL

4.3.1. Manual Calibration

The calibrated parameters were the Horizontal and Vertical Hydraulic Conductivity
corresponding to sand and clay. The evaluation of calibration was based on the error
Variance, Hydraulic head contours and Scatter plots of observed versus simulated
hydraulic head.
Results
The trial and error results for manual calibration are summarized in Table 5, the first
changes were made to KH and Kv for sand, as expected, increased Hydraulic
conductivity resulted in a reduction of Variance. The optimum values are K H=37
[m/day] and Kv=3.7 [m/day]. Afterwards, changes in K H and Kv for clay showed that
the optimum values are KH=2.5[m/day] and Kv=0.25[m/day].
Parameter
s
KH Sand
KH Clay
KV Sand
KV Clay
Variance

Units

Initial

1st

2nd

3rd

4th

5th

[m/day]
[m/day]
[m/day]
[m/day]
[m/day]2

35
3
3.5
0.3
1.33E-03

30
3
3
0.3
1.44E-03

40
3
4
0.3
1.36E-03

37
3
3.7
0.3
1.33E-03

37
3.5
3.7
0.35
1.34E-03

37
2.5
3.7
0.25
1.25E-03

Table 5. Summary of Manual Calibration Trials

The optimum parameters resulted in a Variance of 1.25x10 -3[m/day]2. The hydraulic


head contours and Scatter plot are shown in Figure 10 and Figure 11 respectively. The results
do not show a great variation from the initial values, in fact, the values for the observation
wells are still underestimated.

Figure 10. Hydraulic Head contour after Manual Calibration, Layer 1 (Left), Layer 2 (Center)
and Layer 3 (Right)

Figure 11. Scatter Plot of Observed Vs Simulated Hydraulic Heads after Manual Calibration

4.3.2.Automatic Calibration
The calibrated parameters were Horizontal Hydraulic Conductivity for sand and clay.
The evaluation of automatic calibration was based on statistical indicators such as
Sum of Squared Error SSE, error Variance, Hydraulic head contours and Scatter plots
of observed versus simulated hydraulic head. The time step was equal to 2 days.
Results
The optimum parameters are KH(sand)=30.5 [m/day] and KH(clay)=5, the objective
function reached a value of SSE=0.76. The change in parameter values do not
correlate with the changes made in Manual Calibration and the variation in contour
lines is imperceptible in the same way as for the scatter Plot shown in Figure 12.
In order to improve these results, other parameters as Storage Coefficient were
added to the calibration, however, the estimated parameters lacked physical
meaning and the values were set back to the original.

Figure 12. Scatter Plot of Measured and Simulated Hydraulic Heads

Summary of Calibration
The final hydraulic heads obtained from calibration are presented in Figure 13 and
the values in ANNEX A. The difference between the results is almost imperceptible
and the underestimation at four of the observation boreholes (maximum 50 [cm])
was not removed after automatic calibration. However, considering the values of
residuals and variance, the fit of both manual and automatic calibration is
acceptable.
Table 6 shows a statistics summary of Manual and Automatic Calibration that helped
to select the final best parameters. Considering that the aim is to minimize residual
spread and mean and maximize correlation, it is clear that the results obtained for
Manual Calibration have the best indicators; furthermore, the parameters have
realistic values that represent the study area.

SSE[m ]
Mean Residual [m]
Variance[m2]
R2

Before
0.68
0.197
1.33E-03
0.9935

Manual
0.67
0.191
1.25E-03
0.9936

Table 6 Summary of Statistics for calibration

Automatic
0.76
0.200
1.45E-03
0.9917

Calibration Results for


Boreholes in Layer 1
36
Observed

35 Before Manual Calibration


34

Hydraulic Head [m]

33

After Manual Calibration 32 After Automatic Calibration

31
W1

X1

Y1

Z1

V1

Observation Borehole

Calibration Results for


Boreholes in Layer 3
38
37

Observed

Before Manual Calibration

36
35
Hydraulic Head [m] 34
33
After Manual Calibration

After Automatic Calibration

32
31

A3

B3

C3

D3

E3

Observation Borehole

Figure 13 . Hydraulic Heads for Summary calibration

4.3.3.Validation
Considering the availability of a new set of Hydraulic heads observation for a
different transient period, or steady state period, validation or verification of the

model can be performed by comparing the simulated values with those


measurements. This analysis shows how does the model perform and how well do
the parameters represent the aquifer.
However, for this case there is no available data to validate the model, therefore,
different methods can be applied as described on (De Smedt F., 2012). Among
them, the use of Bracketed ranges consists of carrying one simulation with the
minimum value of the parameters and another with the maximum value, the
obtained range of results contains the real value for the new period.

5. TRANSPORT MODEL
In this Section the calibrated Groundwater Flow model was used to assess the
transport time, direction and concentration of polluted groundwater from an
industrial area into the River or drinking water pumping well.
5.1.

TRANSPORT BY ADVECTION

The tool PMPATH was used to simulate Forward Particle Tracking and determine the
transport time and direction of polluted groundwater with and without pumping. The
industrial area was defined as indicated in Figure 2 and Particles were added in Face
5 (surface), the number NJ=1 and NI=1. The time step was set equal to one year.
Results
As seen in Figure 14, for the case of Transient flow without pumping well, the
particles flowed to the river, the first particles reached there after 11 years and the
last particles after 18 years. For the case of Transient flow with pumping well, the
particles got a little closer to the well but did not reach it. The first particle reached
the river in 12 years and the last particles after 18 years.

Figure 14 Particle Tracking for Transient flow without Pumping well (right) and with
Pumping well (left)

5.2.

TRANSPORT BY ADVECTION AND DISPERSION

The package MT3DMS/SEAWAT was used to calculate the concentration distribution


of pollutants by transport by advection and dispersion, the direction of pollutants
and the transport time with and without pumping well.
MT3DMS/SEAWAT requires specifications, for instance, it was specified that the
solution concentration does not affect the fluid density and that no kinetic reaction
is simulated.
The initial concentration of pollutants was set to 0 [Kg/m3].
Dispersion parameters such as ratio of horizontal transverse dispersivity to
longitudinal dispersivity (TRPT) and ratio of vertical transverse dispersivity to
longitudinal dispersivity (TRPV) were kept equal to the default value 0.1.
Longitudinal dispersivity associated with transport distances and assumed
dependent of scale (Delgado, 2007) was set equal to 10 [m].
The Concentration of pollutants was applied as Recharge over the Industrial Area
and set equal to 1000 [g/m3], this value will remain constant during the analysis.
In order to compare with results obtained with transport by advection, the Total
period of simulation is set equal to 20 years and the Output interval t equal to 365
days in order to avoid excessive computation times.
Results
Figure 15 shows the concentration of pollutants after 20 years of simulation without
pumping well in Layer 1 and Layer 3, which better represent the changes through
the model (Layer 2 provided an intermediate view of Layer 1 and Layer 3).
The direction of pollutant transport can be identified from the shape of the contour
lines for 25 [mg/m3], it reaches the middle of the River in the East boundary. At the
well location (1450, 550), observations from a borehole confirm that the
concentration after 20 years is almost null, see Figure 16.
The concentration of pollutant that reaches the river after 20 years is lower than 17
[mg/m3] and the extent of the contribution is approximately 600 m.
The process of dispersion is reflected in the extent of the area polluted, and also in
the difference in concentration gradient between Layer 1 and Layer 3. The
concentration in Layer 1 is higher because the contaminant travels a minor distance
along the depth of the aquifer and thus, disperses less in that direction. That
explains why the contaminant in Layer 3 (depth of 40 m) reaches the East boundary
with a concentration of lower than 9 [mg/m3].
We can also see the gradient of dispersion in the Industrial area, the direction of
groundwater flow from high to low hydraulic head resulted in an accumulation of
pollution in the west side of the industrial area in Layer 1.

Figure 15 Pollutant concentration for Layer 1 (left), Layer 3 (right) for the Case without
Pumping Well

Figure 16 Pollutant concentrations observed in Well after 20 years without pumping

Figure 17 shows the concentration of pollution after a 20 year period with pumping
well in Layer 1 and Layer 3. If the well pumps during a 20 year period, the direction
of the contaminated flow deviates in the direction of the well, but the concentration
in that area is lower than 2,5 [mg/m3] as shown in Figure 18.
However, most of the pollution still flows to the river but this time, the affected part
extends 600 m towards the most downstream part of the river and the maximum
concentration is approximately 9 [mg/m3], the reduction in respect to the case
without well is clearly explained by the fact that the well is pumping out part of the
contaminant.

Figure 17. Pollutant concentration for Layer 1 (left) and Layer 3 (right) for the Case with
Pumping Well

Figure 18 Pollutant concentrations observed in Well after 20 years with pumping

5.3.

EFFECT OF EFFECTIVE POROSITY AND DISPERSION COEFFICIENT

In this Section we will evaluate the effect of change in Longitudinal Dispersivity [m]
and Effective Porosity in polluted flow direction, concentration after a determined
time and extension of contaminated area.
We will compare the results for an increase of Longitudinal Dispersivity from 10 [m]
to 50 [m] and for an increase of Effective Porosity of 0.25 (sand) and 0.01 (clay) to
0.5(sand) and 0.05 (clay). The evaluation will be performed in the condition of
pumping well and for Layer 1.
Results
a) Increase in Longitudinal Dispersivity from 10 [m] to 50 [m] (n e as
original)

Figure 19 shows the Concentration contour maps and the Concentration-Time


Curves for different values of Longitudinal Dispersivity, we can see that for higher D L
the extent of the polluted area has increased considerably. Indeed, D L approximates
the spreading of the concentration of pollution and this explains the increase in the
area.
In order to assess the effect of DL in the change of concentration in respect to time
the concentration in the industrial area was compared for both cases. After 20 years
of pollution, for a DL=10 [m] the maximum concentration is approximately 33
[mg/m3] and for a DL =50 [m] the maximum concentration reduces to
approximately 25 [mg/m3], this decrease is explained by the distribution of
pollutant over a larger area.
In respect to the transport time to the river, it is evident that polluted flow with
higher concentration flows slower for a higher D L, see the area with 25[mg/m3]. This
is explained by the inverse relation that exists between D L and vx indicated in
Equation 4Equation 4

Figure 19 Pollutant concentration for Layer 1 (with pumping well) for D L=10 [m] (left) and
50 [m] (right) after 20 years

b) Increase of Effective Porosity of 0.25 (sand) and 0.01 (clay) to


0.5(sand) and 0.05 (clay). (DL=10 [m])
Figure 20 shows the Concentration contour maps and the Concentration-Time
Curves for different values of Effective Porosity, higher n e does not influence the
extent of the polluted area significantly nor the direction.
The Concentration-Time curves for the Observation Borehole V1 show that after 20
years in the aquifer with lower ne the concentration reaches 2.4 [mg/m3], in
contrast, for higher ne the concentration is only to 0.08 [mg/m3]. The same trend is
identified in the gradient of concentration contour lines indicating a smaller
dispersion of pollution for higher n e. For the same condition, it is evident at the river

boundary that the polluted area has reached with a minimal concentration along a
small distance of 50 m. This is explained by the fact that velocity of groundwater
varies inversely according to the size of the pores, an increase in n e means that the
effective area through which groundwater flows is larger and therefore velocity is
lower.

Figure 20 Pollutant concentration and Concentration-Time curve at V1 for Layer 1 (with


pumping well)
for n e=0.25 (sand) and 0.01 (clay) (up) and
ne=0.5 (sand) and 0.05 (clay) (down)

6. CONCLUSIONS
This study aimed to assess the contamination of groundwater in an aquifer by the
construction and analysis of a groundwater flow and a pollutant transport model.

The groundwater flow model was built and afterwards, hydraulic conductivities were
calibrated following manual and automatic approach. The statistical and graphical
evaluation (Var=1.25x10-3 [m2], R2=0.99, ME=0.19 [m]) suggested that the
parameters obtained with manual calibration represented the physical
characteristics of the aquifer in a better way.
The analysis of aquifer contamination was intended to assess the behavior of
pollutant transport in conditions with and without pumping well, when advection
and advection-dispersion are simulated and, when longitudinal dispersion and
effective porosity vary. The analyzed characteristics of polluted transport flow were
direction, transport time, extent and concentration. The results showed that the
direction of contaminated flow reaches the well (with a concentration of 2.4
[mg/m3]) only when diffusion is accounted for and after a period of 20 years, in the
other cases the river receives all contaminated water. In respect to transport time, it
was determined with PMPATH (only advection) that polluted particles reach the river
in a lapse of approximately 7 years after 11 years of the emission, on the other
hand, the results obtained with MT3MTS (advection and diffusion) showed that after
20 years of emission we expect that polluted flow reaches the river with a maximum
concentration of 17 [mg/m 3]. It was found that an increase of effective porosity and
longitudinal dispersivity lead to higher transport times and that the extent of
polluted groundwater area depends mainly on longitudinal dispersivity. Finally, the
concentration of pollutants decreases as transport time increases.

7. REFERENCES
B.P. Leonard, 1988. Universal Limiter for Transient Interpolation Modeling of the
Advective Transport Equations: The ULTIMATE Conservative Difference
Scheme (No. 1009 16). Institute for Computational Mechanics in Propulsion,
Lewis Research Center, Cleveland, Ohio.
Delgado, J.M.P.Q., 2007. Longitudinal and Transverse Dispersion in Porous Media.
Chem. Eng. Res. Des. 85, 12451252. doi:10.1205/cherd07017
De Smedt F., 2012. GROUNDWATER MODELLING Course Notes.
Johnson, A.I., 1967. Specific yield: compilation of specific yields for various
materials. US Government Printing Office.
Simcore Software, 2012. Manual Processing Modflow.
Stephens, D.B., Hsu, K.-C., Prieksat, M.A., Ankeny, M.D., Blandford, N., Roth, T.L.,
Kelsey, J.A., Whitworth, J.R., 1998. A comparison of estimated and calculated
effective porosity. Hydrogeol. J. 6, 156165.
Thangarajan M., 2007. Groundwater: Resource Evaluation, Augmentation,
Contamination, Restoration, Modeling and Management. Springer, India.

8. ANNEX A

Well
W1
X1
Y1
Z1
V1

Observ
ed
35.2
34.7
33.8
32.1
33

Hydraulic Head [m]


Before
Manual
Calibration
Calibration
Calculat Residu Calculat Residu
ed
al
ed
al
35.034
0.166
35.031
0.169
34.703 -0.003 34.703 -0.003
33.429
0.371
33.460
0.340
32.061
0.039
32.081
0.019
32.693
0.307
32.729
0.271

Automatic
Calibration
Calculat Residual
ed
35.037
0.163
34.689
0.011
33.374
0.426
32.082
0.018
32.687
0.313

Summary of calibration for Observation Boreholes in Layer 1

Well
A3
B3
C3

Observ
ed
35.6
34.9
37.2

Hydraulic
Before
Calibration
Calculat Residu
ed
al
35.502
0.098
34.751
0.149
36.877
0.323

Head [m]
Manual
Calibration
Calculat Residu
ed
al
35.490
0.110
34.751
0.149
36.826
0.374

Automatic
Calibration
Calculat Residual
ed
35.558
0.042
34.718
0.182
36.924
0.276

D3
E3

32.3
34.2

32.314
33.672

-0.014
0.528

32.339
33.678

-0.039
0.522

32.324
33.612

-0.024
0.588

Summary of calibration for Observation Boreholes in Layer 3

Vous aimerez peut-être aussi