Linked simulation-optimization based methodologies for unknown groundwater pollutant source identification in managed and unmanaged contaminated sites

Jha, Manish Kumar (2012) Linked simulation-optimization based methodologies for unknown groundwater pollutant source identification in managed and unmanaged contaminated sites. PhD thesis, James Cook University.

PDF (Thesis)
Download (10MB) | Preview
View at Publisher Website:


Groundwater is the primary source for irrigation and drinking in many parts of the world. Anthropogenic activities such as mining; large scale production, storage and transport of various chemicals; improper waste management practices; and unsustainable intensive agricultural practices have resulted in the contamination of many groundwater aquifers. The identification of the exact location and release history of contributing sources, which are often unknown, is very important in planning effective remediation measures as well as in determining the liability on the polluter. Contamination of ground water aquifers may be caused by a combination of pollutant sources varying in time of release, flux and location. In situations where they are unknown, location and release histories have to be estimated by inversion. Inversion of the equations governing flow and transport over time and space is an ill-posed problem.

Estimation of unknown groundwater pollutant source characteristics from measured pollutant concentrations at several monitoring locations is generally an ill-posed and sometimes non-unique inverse problem. Linked simulation-optimization based methodologies have evolved as effective tools capable of solving this problem.

One of the important issues in estimation of unknown contamination sources is the release history reconstruction. It is generally assumed that reliable information on potential source locations and their time of activity is available from background studies of anthropogenic activities on a contaminated site. In such cases, only the release history of the pollutant sources is unknown. Some of the main limitations in accurate source characterization are:

1. Sparsity of concentration measurement data. 2. Inefficient monitoring network for concentration measurements. 3. Difficulty in establishing the time of pollutant source activity initiation. 4. Applicability of optimal source characterization to distributed sources. 5. Problems associated with achieving a global optimal solution efficiently.

In order to address some of these limitations, initially a linked simulation-optimization model for optimal source characterization is developed using adaptive simulated annealing (ASA) as the optimization algorithm. Performance of the ASA based methodology was compared with a source characterization method using genetic algorithm (GA) for optimization, in terms of their ability to handle uncertainties and efficiency of convergence. Using illustrative aquifer examples, it was shown that ASA converges faster and produces better results even with erroneous measurement data and with uncertainties in hydraulic conductivity and porosity. A more complex scenario exists when no reliable information is available on the potential location or initial time of activity of sources. Apart from this, the frequency of measurement at monitoring wells may not be uniform and some measurements might be missing in practical situations.

A methodology is developed to generate initial estimates of source characteristics such as source location and to estimate the initial time of activity from pollutant concentration measurements obtained from a single location where the contamination was first detected. dynamic time warping (DTW) distance is used to minimize errors in estimation of source characteristics arising from improper alignment of estimated and observed concentration data on the temporal axis. Performance of this methodology is evaluated using data obtained from both an illustrative site and an actual contaminated site. Based on these estimates, a methodology is developed to design a monitoring network to generate concentration measurement information aimed at obtaining more reliable estimates of source characteristics. This methodology is implemented for a real contaminated site and it was found that the use of developed methodology results in reliable estimates of source characteristics with a far lesser number of monitoring wells.

The source characterization methodology is then extended for estimation of release history of distributed pollutant sources in a realistic scenario. Distributed sources in an abandoned mine site were considered for this purpose. A conceptual flow model is developed and calibrated for an abandoned mine site in South-East Queensland. Various illustrative scenarios of contamination are considered for evaluating the performance of this developed methodology. It was shown that the developed methodology is potentially applicable for estimation of distributed source characteristics.

When management measures are implemented to control contamination in a groundwater aquifer, measured concentration values are the resultant effect of natural transport and control measures. This can produce incorrect estimates for source characteristics. The methodology for release history reconstruction can be applied to managed contaminated sites by incorporating the proposed management strategy into the groundwater flow or transport model. This is illustrated by incorporating contamination management strategies already in place into the groundwater flow and transport model for an abandoned mine site with some degree of existing control measures.

Item ID: 40015
Item Type: Thesis (PhD)
Keywords: annealing; environmental engineering; groundwater; hydrogeology; hydrology; mathematical optimization; pollution; simulated annealing; simulation methods; simulation-optimization; water management; water resources engineering; water resources
Related URLs:
Additional Information:

For this thesis, Manish Kumar Jha received the Dean's Award for Excellence 2014.

Publications arising from this thesis are available from the Related URLs field. The publications are:

Chapter 3: Jha, Manish, and Datta, Bithin (2013) Three-dimensional groundwater contamination source identification using adaptive simulated annealing. Journal of Hydrologic Engineering, 18 (3). pp. 307-317.

Jha, Manish Kumar, and Datta, Bithin (2014) Linked simulation-optimization based dedicated monitoring network design for unknown pollutant source identification using dynamic time warping distance. Water Resources Management, 28 (12). pp. 4161-4182.

Jha, Manish, and Datta, Bithin (2012) Application of simulated annealing in water resources management: optional solution of groundwater contamination source characterization problem and monitoring network design problems. In: Sales Guera Tsuzuki, Marcos de, (ed.) Simulated Annealing- Single and Multiple Objective Problems. InTech, Rijeka, Croatia, pp. 157-174.

Jha, Manish K., and Datta, Bithin (2011) Simulated annealing based simulation-optimization approach for identification of unknown contaminant sources in groundwater aquifers. Desalination and Water Treatment, 32 (1-3). pp. 79-85.

Date Deposited: 12 Aug 2015 06:52
FoR Codes: 09 ENGINEERING > 0905 Civil Engineering > 090509 Water Resources Engineering @ 50%
09 ENGINEERING > 0907 Environmental Engineering > 090799 Environmental Engineering not elsewhere classified @ 50%
SEO Codes: 96 ENVIRONMENT > 9609 Land and Water Management > 960999 Land and Water Management of Environments not elsewhere classified @ 100%
Downloads: Total: 219
Last 12 Months: 4
More Statistics

Actions (Repository Staff Only)

Item Control Page Item Control Page