Simulation of complex geochemical processes in contaminated aquifer sites using surrogate models as approximate simulators

Esfahani, Hamed, and Datta, Bithin (2015) Simulation of complex geochemical processes in contaminated aquifer sites using surrogate models as approximate simulators. In: Abstracts from the 6th International Contaminated Site Remediation Conference. P5. pp. 456-457. From: CleanUp 2015: 6th International Contaminated Site Remediation Conference, 13-16 September 2015, Melbourne, VIC, Australia.

[img] PDF (Extended abstract) - Published Version
Restricted to Repository staff only

View at Publisher Website:


Reactive transport processes of AMD, in contaminated groundwater system, especially with multiple species, is a complex and highly non-linear process. Simulation of such complex geochemical processes using efficient numerical models is generally computationally intensive. In many contaminated groundwater aquifer management scenarios, an efficient strategy is necessary for effective and reliable remediation and control of the contaminated aquifer. Also, in a hydrogeologically complex aquifer site e.g., mining site, acid mine drainage (AMD) and the reactive chemical species together with very complex geology complicates the characterization of contamination source location and pathways. In such contamination cases, it becomes necessary to develop optimal source characterization models, and strategies for future remediation. Solution of optimization models either for source characterization or optimal management strategy development requires the incorporation of the complex physical processes in the aquifer (Sreekanth & Datta, 2011). Also, most of the developed optimization models for source characterization or remediation strategy development require repeated solution of the numerical simulation models within the optimization algorithm. This process is enormously time consuming and often restricts the computational feasibility of such optimization approaches (Sreekanth & Datta, 2011). In order to overcome these computational restrictions, and to ensure computational feasibility of characterizing sources and pathways of contamination it is computationally advantageous to develop surrogate models which can be trained using solutions obtained from rigorous numerical simulation models. Trained and tested ensemble Genetic Programming (GP) models are developed as robust surrogate models, using the simulated response of a complex contaminated aquifer to randomly generated source fluxes (Datta et al., 2013). Performance of GP models are evaluated by comparing GP model solutions with solution results obtained by using a rigorous numerical simulator, HYDROGEOCHEM 5.0 [HGCH] (Sun, 2004), to simulate flow and transport processes for a study area resembling an abandoned mine site incorporating acid mine drainage. Therefore, this study develops ensemble GP models to approximately simulate 3-dimensional, reactive, multiple chemical species transport in contaminated aquifers.

Item ID: 43797
Item Type: Conference Item (Research - E1)
ISBN: 978-1-921431-47-0
Date Deposited: 22 Apr 2016 02:06
FoR Codes: 09 ENGINEERING > 0905 Civil Engineering > 090509 Water Resources Engineering @ 100%
SEO Codes: 96 ENVIRONMENT > 9609 Land and Water Management > 960999 Land and Water Management of Environments not elsewhere classified @ 100%
Downloads: Total: 3
More Statistics

Actions (Repository Staff Only)

Item Control Page Item Control Page