Simulation of subsurface water flow by galerkin finite element method in the Dhaka city aquifer

Abstract

A two-dimensional mathematical model has been developed USll1gGalerkin finite element method for simulation of transient saturated-unsaturated groundwater flow in multiaquifer system with window based, menu-driven, user friendly interface to provide on screen input data interpretation, to guide user through steps of data processing, and to generate reports and graphs from output results. The model is able to handle combined and system-dependent boundary conditions that can be used in practical situation of groundwater flow in the complex hydrogeological conditions of a country like Bangladesh in general and particularly for Dhaka City. Due to less complexity in 2-D simulation and the availability of GUI the model is easier to use and time efficient. The p.(ogram deals with a variety of boundary conditions encountered in real hydrogeological situation consisting of simple constant or time dependent prescribed head (surface water bodies) or prescribed flux boundaries (pumping well, recharging well etc.), complicated system-dependent and atmosphere-controlled boundary including the near-surface activities i.e., precipitation, evaporation, root water extraction and surface ponding caused by excess precipitation and also boundary along the riverside having variable seepage face conditions. With a view to make the model more user friendly and efficient in operation a Graphical User Interface (GUI) is designed to handle the user input, reporting the output and graph generation. This GUI is designed and implemented using Object Oriented Programming language Visual Basic that will make the Fortran program operable in the 32-bit faster windows environment. To test the different features incorporated in the model, a number of published schematized problems have been simulated by the program. Five such problems are presented in detail in this thesis with simulated results. The program and the modification made were tested on the following examples: I. One-dimensional vertical soil column with the variable atmospheric boundary condition and with water extraction by plant roots, 2. One-dimensional vertical soil column without the variable atmospheric boundary condition 3. two-dimensional rectangular vertical flow domain with the variable seepage-face boundary condition, without influence of atmosphere. 4. Two-dimensional vertical flow domain representing a simplified nver bank situation with both variable atmospheric and variable seepage-face boundary conditions (Coarser Grid) 5. Two-dimensional vertical flow domain representing a simplified nver bank situation with both variable atmospheric and variable seepage-face boundary conditions (Finer Grid) The effect of grid fineness is tested in examples 4 and 5. To verify the model responses in simulating a complex real hydrogeological situation Dhaka City aquifer is studied. The boundary conditions of the domain are derived from the river water level and piezometric level data. Model calibration of the parameters has been performed under this study. Simulated water balance components identify the river effect as the most dominating recharge mechanism of the area, which contributes around 60 percent of the total volume of abstraction. Urban recharge (taken 30% of the total abstraction) also contributes to the abstraction volume of the area. The results of simulation show that the approach is fruitful and can be used for detailed study of the groundwater status of the Dhaka City aquifer in future. The program is proved to work well but have a disadvantage of being associated with the time step of the atmospheric input data, which is usually I day. Shorter and/or automatically set time step for the atmospheric input data, as well as for the boundarycondition subroutines, is therefore recommended. A need has been also identified for yet more flexible treatment of combined and system-dependent boundary conditions (e.g. a combination of the variable atmospheric boundary condition with the variable seepage-face one) and for the coupling of the present model with a two-dimensional model of surface runoff in a flood plain.