Yenigul, N.B., Elfeki, A.M.M. (2002). Influence of Subsurface Heterogeneity on Detection of Landfill Leakage. Published at 14th International Conference Computational Methods in Water Resources “CMWR2002”, Developments in Water Science No. 47, Elsevier Publisher, The Netherlands, pp. 1323-1330. Eds. SM Hassanizadeh, RJ Schotting, WG Gray, and GF Pinder.

1. Optimum Design of Groundwater
Monitoring Networks at Landfill
Sites
Nusin Buket Yenigul
Prof. Dr. C. van den Akker
Dr. A.Elfeki
Dr. J.C.Gehrels
Faculty of Civil Engineering & Geosciences
Department of Hydrology and Ecology

2. Content
Research Outline
 Influence Of Uncertainty In Leak Location On Detection
of Contaminant Plumes Released At Landfill Sites
Objectives
Hypothetical Test Cases
Results of the analysis
Motivation and Objectives
 Influence Of Subsurface Heterogeneity On Detection of
Landfill Leakage
Objectives
Hypothetical Test Cases
Results of the analysis
 Concluding Remarks
 Future Plan

3. Formulation of a methodology for the design
of an optimum monitoring well network at a
landfill site.
Motivation and Objectives
Highest probability of
contaminant detection
Cost effectiveEarly detection

4. Research Outline
Effects due to spatial heterogeneity of the subsurface
GROUNDWATER FLOW AND TRANSPORT MODEL
STOCHASTIC CHARACTERIZATION & SENSITIVITY ANALYSIS
Influences related to the uncertainties in contaminant source location
Steady state uniform flow
Transient flow
Random walk transport model
Influence of number of wells, on the detection probability
Influence of dimension of the source & detection limit on the detection probability
Influence of dispersivity of medium on the detection probability
Influence of pumping & sampling frequency on the detection probability
OPTIMIZATION
trade-off among the maximum detection probability, early detection and minimum cost.
APPLICATION OF METHODOLOGY
Application to a real case study.
FORMULATION OF GUIDELINES

5. Cooperation With
TNO
GEODELFT
TAUW
TU DELFT MATHEMATICS DEPARTMENT
Publication
Influence of Uncertainty In leak Location On Detection of
Contaminant Plumes Released at Landfill Sites
Modelcare 2002, 4th International Conference on Calibration And Reliability In
Groundwater Modelling, Praque, Czech Republic, 17-20 June 2002”
Influence of Subsurface Heterogeneity on Detection of Landfill
Leakage
CMWR 2002, 14th International Conference on Computational Methods in Water
Resources, Delft, The Netherlands, 23-28 June 2002”

6. Influence Of Uncertainty In Leak Location
On Detection Of Contaminant Plumes
Released At Landfill Sites
“Presented in Modelcare 2002”

7.  uncertainties due to subsurface heterogeneity
Objectives
To Analyze The Influence Of :
 uncertainties due to contaminant leak location
 dispersivity of medium
 number of wells in monitoring system
 the initial contaminant source size

8. 0 20 40 60 80 100 120 140 160 180 200
-200
-180
-160
-140
-120
-100
-80
-60
-40
-20
0
M-1
M-2
M-3
M-4
M-5
M-6
M-7
M-8
M-9
M-10
Landfill
Flow direction
Plan View of Model Domain

9.  Steady state groundwater flow
 2000 particles with a total mass of 1000 gram
 Zero flux and constant head
 Hydraulic gradient is 0.001
 Confined aquifer
 Y= ln (K) is modeled as a Gausian stationary
distribution
 2
Y is set to “0”, “1” and “2” and x= x =5 m
 Monte Carlo method is used to generate leak locations
Hypothetical Test Model

10.  Random leak locations follow a uniform distribution
 Failure is modeled as a point and a small areal source
 Detection limit corresponds the detection of the first
particle hits the well
 L= 0 m, T= 0 m (advection); L= 0.5 m, T= 0.15 m;
L= 1.5 m T= 0.15 m
 porosity = 0.25
 contaminant are assumed to be conservative
Hypothetical Test Model

11. 0
5
10
15
20
25
30
35
40
0 1 2 3 4 5 6 7 8 9 10 11
number of the wells
detectionprobability(%)
0
1
2
L=0T=0
x=y= 5 m
2
Y=
Influence of 2
Y On Monitoring Systems of 3, 5
& 10 wells for Point Contaminant Source

12. 0
5
10
15
20
25
30
35
40
0 1 2 3 4 5 6 7 8 9 10 11
number of the wells
detectionprobability(%)
0
1
2
L=0T=0
x=y= 5 m
2
Y=
Influence of 2
Y On Monitoring Systems of 3, 5
& 10 wells for Areal Contaminant Source

13. 0
10
20
30
40
50
60
70
80
90
100
0 1 2 3 4 5 6 7 8 9 10 11
number of the wells
detectionprobability(%)
L=0,T=0
L=0.5,T=0.05
L=1.5,T=0.15
Influence of Dispersivity On Monitoring Systems of
3, 5 & 10 Wells for Areal Contaminant Source
(2
Y=0)

14. Subsurface heterogeneity detection probability
Number of wells detection probability
Dispersivity of medium detection probability
Current practice (3 wells) is not sufficient.
Initial size contamination source detection probability
Results of The Analysis

15. Influence Of Subsurface Heterogeneity On
Detection Of Landfill Leakage
“Presented in CMWR 2002”

16.  To analyze the spatial variability of hydraulic
conductivity on contaminant plume detection
Purpose
 To characterize the subsurface heterogeneity based on
Gaussian and Non-gaussian models
 The comparison of the results of two approaches

17.  Hydraulic conductivity is assumed to be the major contributor to
the uncertainty
 Logarithm of hydraulic conductivity (ln K) is modeled;
1) as a Gaussian stationary distribution with mean, variance and a
correlation length,
2) as a non-Gaussian distribution using a coupled Markov chain
model (CMCM).
 A Monte Carlo method is used to generate multiple random hydraulic
conductivity field.
 Steady state groundwater flow model
 random walk transport model
 Contaminants are assumed to be conservative.
 L=0 m, T=0 m; L=0.5 m, T=0.05 m; L=1.5 m, T=0.15 m.
 4 geological units are considered in coupled CMCM
Hypothetical Test Model

18. 0 20 40 60 80 100 120 140 160 180 200
-200
-180
-160
-140
-120
-100
-80
-60
-40
-20
0
1
2
3
4
Units
Geological
Flow direction
Landfill
Leakage
MW 1
MW 2
MW 3
MW 4
MW 5
Plan View of Geological Sample

19. Unit
Color in
Figure 1
Wi Low Contrast High contrast
1 yellow 0.24 80 m/day 100 m/day
2 blue 0.25 50 m/day 10 m/day
3 red 0.31 20 m/day 1 m/day
4 green 0.20 10 m/day 0.1 m/day
Parameter Low Contrast High Contrast
Km(m/day) 39.9 26.8
K 26.7 41.2
Y=lnK 3.5 2.68
Y 0.61 1.1
x 25.0 m 25.0 m
y 2.0 m 2.0 m
Hydraulic conductivity values of the units in non-Gaussian (Markovian) field.
Estimated simulation parameters for generation of statistically equivalent
Gaussian fields.

20. 0 20 40 60 80 100 120 140 160 180 200
-200
-180
-160
-140
-120
-100
-80
-60
-40
-20
0
0
10
20
50
80
100
200
300
400
K
(m/day)
Gaussian conductivity field
with low contrast.
Non-Gaussian conductivity field
with low contrast.
0 20 40 60 80 100 120 140 160 180 200
-200
-180
-160
-140
-120
-100
-80
-60
-40
-20
0
10
20
50
80
K
(m/day)

21. 0
10
20
30
40
50
60
70
80
90
100
advection dispersivity=0.5 dispersivity=1.5
mw1
mw2
mw3
mw4
mw5
detectionprobability(%)
Detection Probabilities of Monitoring Wells in Low Contrast
Non-gaussian (Markovian) Case

22. Detection Probabilities of Monitoring Wells in Low Contrast
Gaussian Case.
0
10
20
30
40
50
60
70
80
90
100
advection dispersivity=0.5 dispersivity=1.5
mw1
mw2
mw3
mw4
mw5
detectionprobability(%)

23. Results of The Analysis
 Detection probabilities in non-Gaussian and Gaussian
cases are slightly different.
Less discrete variation Gaussian stationary distribution.
Complex geology with particular features Markov model
Dispersivity of medium detection probability

24. Concluding Remarks
Detection probability of contaminant plumes highly
depends on:
subsurface heterogeneity
size of the plume
number of the wells in a monitoring system
Efficiency of 3 well system particularly in medium with
relatively low dispersivity is quite dubious
in case of less discrete variation between the
geological units, subsurface heterogeneity can be
modeled based on a Gaussian stationary distribution.

25. Future Plan of Work (2003)
Continue Calculations for Stochastic Characterization
and Sensitivity Analysis
• To create test models representing hydrogeological conditions in
east and west part of The Netherlands
• Designing of various monitoring networks to be utilized in
formulation of guidelines
• Developing an analytical approach that can provide compatible
results with the simulation model
• Analyzing the detection probability of each network to be used
in optimization model in far steps of the research
Literature study
Publications