-
author = "Om Prakash",
-
title = "Optimal monitoring network design and identification
of unknown pollutant sources in polluted aquifers",
-
school = "School of Engineering and Physical Sciences, James
Cook University",
-
year = "2014",
-
address = "Australia",
-
month = apr,
-
keywords = "genetic algorithms, genetic programming",
-
URL = "http://researchonline.jcu.edu.au/37017/",
-
URL = "http://researchonline.jcu.edu.au/37017/1/37017-prakash-2014-thesis.pdf",
-
size = "229 pages",
-
abstract = "Increasing stress from various anthropogenic
activities has resulted in widespread pollution of
groundwater resources. Often, when the pollutant is
first detected in groundwater, little is known about
the pollutant sources. Identification of source
characteristics in terms of locations, activity
initiation times, and source flux release histories and
activity durations are vital in planning effective
remediation measures and determining the liability of
the polluter. Groundwater pollution source
characterization is an inverse and ill-posed problem.
Finding a solution to this inverse problem remains a
challenging task due to uncertainties in accurately
predicting the aquifer response to source flux
injection, generally encountered sparsity of
concentration measurements in the field, and the
non-uniqueness in the aquifer response to the subjected
hydraulic and chemical stresses. This study presents
linked simulation-optimization, and sequential
monitoring network design based methodologies for
identification of unknown groundwater pollution source
characteristics.",
-
absttract = "Pollution in groundwater aquifers is generally first
detected in an arbitrarily located water supply well or
a group of wells. Often pollutants are detected much
after activity at the sources may have initiated, or
even after it has ceased to exist. There may be a gap
of years, or even decades, between the start of source
activity and detection of pollutants in an aquifer.
Other important issues in accurately identifying
unknown groundwater pollution source characteristics
are the quality, usability and extent of pollution
measurement data from the study area. Existing
methodologies for unknown groundwater pollution source
characterization have several limitations.
Methodologies developed in this study aim to address
some of these limitations. The major limitations
addressed in this study include:
i. sparsity of pollutant concentration measurement
data,
ii. inefficient monitoring network for concentration
measurements,
iii. difficulty in identifying the source
locations,
iv. difficulty in establishing the pollutant source
activity initiation time,
v. applicability of optimal source characterization
with missing observation data.",
-
absttract = "In many cases of aquifer pollution, especially in
clandestine underground disposal of toxic wastes, no
information is available about the number and location
of such sources. Moreover, monitoring wells where
pollution is first detected may not be optimally
located for accurately identifying the release history
of unknown pollution sources. A large number of
pollutant concentration measurements spread over time
and space is necessary for accurate source
identification. However, long term monitoring over a
large number of monitoring locations has budgetary
constraints. This study presents a sequential optimal
monitoring network design methodology based on
geostatistical kriging, a pollutant concentration
gradient based search for identification of source
locations, and a Genetic Programming (GP) based optimal
monitoring network design model for collecting
concentration measurements for efficient source
characterization.
To address the issue of unknown starting times of
activity of the sources, a new methodology is developed
for simultaneously identifying the starting times of
the activity of the sources and their flux release
history. A new optimum decision model is formulated and
solved such that the starting times of the activity of
the sources are directly obtained as solution.
Simulated Annealing (SA) is used for solving the
optimization problem with the starting time of
pollutant source activity incorporated as explicit
decision variable.
Subsequent to the detection of pollution in an aquifer,
a more formal methodology for source characterization
is generally initiated only after large numbers of
spatiotemporal concentration measurements, spaced over
a sufficiently long period of time, are obtained.
During this time, the spread of the pollutant continues
while temporal measurements are being obtained at
monitoring locations. A feedback-based sequential
methodology for efficient identification of unknown
pollutant source characteristics, integrating optimal
monitoring network design and an optimization based
source identification model, is developed. The main
advantage of this methodology is that source
characterization can start at the same time as when
pollutant is first detected in the aquifer. In every
sequence, feedback from the source identification model
improves the optimal monitoring network design and
vice-versa. This results in efficient and accurate
source characterization, within a few sequences of
source identification and monitoring network design.",
-
absttract = "The performances of the developed methodologies are
evaluated for different scenarios of groundwater
pollution incorporating transient flow and
advective-dispersive transport in heterogeneous
anisotropic conditions. The applicability of the
developed methodologies is tested for a real aquifer
site polluted with petrochemical waste (BTEX). These
evaluation results demonstrate the potential
applicability of the developed methodologies to
correctly estimate the unknown source flux's magnitude,
and location and source activity initiation times,
while improving the accuracy of source flux
identification. Results of performance evaluation of
each of these methodologies indicate their potential
for field application.",
-
notes = "Genetic Programming Models for Impact Factor
Assessment and Frequency Factor Assessment
Item ID: 37017 Supervisor Bithin Datta",