OPTIMAL IDENTIFICATION OF PARAMETERS AND SIMULATION OF GROUNDWATER FLOW

Abstract

The present dissertation embodies research work pertaining to two broad categories of groundwater flow problems, i.e. identification of parameters and simulation of groundwater flow. The first category, i.e. the identification problems, researched and presented in this dissertation include: identification of aquifer parameters i.e. Storage coefficient, S, and Transmissivity, T; and identification of well hydraulic parameters i.e. the formation loss coefficient, entry loss coefficient and turbulence loss coefficient. In the first attempt, a formulation is developed in framework of Genetic Algorithm (GA) and a computer model is developed, which inversely and optimally identifies aquifer parameters, Transmissivity and Storativity. The model and its formulation are validated by comparing the model results with the ones obtained through the Theis conventional approach and also with the prototype data. Minimization of the objective function logically convinces towards substantiation of the solution obtained in the framework of GA technique. Next, an optimal identification of the well hydraulic parameters, i.e., formation loss coefficient, entry loss coefficient, turbulence loss coefficient, and eventually determination of well efficiency, is achieved. Jacob’s equation is formulated in the framework of GA, which ultimately resulted in the development of a computer model coded in C ++ language. The parameters are identified through minimization of an objective function. Moreover, validity of the model is also endorsed on the basis of xixstatistical analysis while comparing simulated drawdowns against the observed ones. Through the present research, Jacob and Singh’s assumptions for constant value of turbulence loss coefficient is found misleading; whereas, Rorabaugh and Todd’s opinion is ascertained. Hence, it may be concluded that, the formulation made in the sense of GA technique and the computer model developed in this study, are valid enough to use. Under the second category, i.e. simulation of groundwater flow, firstly a two- dimensional simulation of steady-state groundwater flow in an unconfined aquifer is carried out. A computer model in the framework of FEM is developed, coded in FORTRAN language, and applied to a real field problem consisting of two- dimensional horizontal domain for prediction of piezometric heads, flow field under the most drought period of the year. The results are substantiated on the basis of minimization of residuals between observed and simulated piezometric heads at the Dirichlet points. Furthermore, the model estimates the effective lift and accordingly a tubewell zoning map is prepared, which is deemed to be a good tool for guidance of irrigation practitioners to opt for appropriate tubewell technology by signifying the proper locations. Lastly, a two-dimensional equation of mass conservation; governing saturated groundwater flow under transient conditions in a confined aquifer is simulated with source or sink terms by developing a computer model, coded in FORTRAN language, in the framework of FEM. While employing FEM in spatial sense, weighted residual technique is used as per Galerkin’s approach; whereas time derivative is solved by using interpolation functions and lumped element type scheme of Crank-Nicholson. xxThe model is implemented to a real field problem consisting of two-dimensional horizontal domain together with a sink penetrating fully into a confined aquifer for prediction of piezometric heads (equipotential lines) and thereby by formation of cone of depression, distribution of flow field signifying stream lines with velocity vectors and determination of the radius of influence of the well when it achieves stationary condition. The simulated results are substantiated on the basis of minimization of residuals between observed and simulated piezometric heads. Performance of the model, tested statistically, rank the model very satisfactory.