MODELING AND DESIGN OF SCAVENGING SKIMMING WELL SYSTEM IN FRESH-SALINE AQUIFERS

Abstract

The existence of freshwater overlying saline water in groundwater system is widespread in many inland aquifers as well as in the coastal aquifers. The withdrawal of freshwater overlying saline groundwater results in saline water upconing leading to inferior quality of pumped water and degradation of the aquifer. The continuous use of such inferior quality water for irrigation adversely affects fertile lands causing agro-economic declination. In such situations, skimming wells can play an important role in augmenting irrigation supplies as well as controlling water logging and soil salinity problems. Traditionally, skimming wells have been partially penetrating wells tapping only the upper portion of the freshwater lens, and keeping acceptable vertical distance between the screen bottom and the interface to accommodate the upconing during pumping. The performance of the partially penetrating skimming wells in shallow fresh-saline aquifers may be severely jeopardized because of relatively quicker rise (upconing) of the underlying saline water. The problem of quick upconing can be addressed by providing a compound well system comprising closely spaced multiple screens viz. scavenger and recirculation well systems. The scavenger well system consists of two pumping wells viz. the production well and an additional scavenger well. The production well taps the freshwater zone while the scavenger well taps the saline water zone. Simultaneous pumpage from the scavenger well attenuates the upconing of the interface, and hence improves the quality of the water emanating from the production well. The recirculation well system also consists of two wells viz. the production well and an additional recharging well (termed as recirculation well and is located below the production well) - both located in the freshwater zone. The production well taps the freshwater zone and a portion of the pumped freshwater is injected back into the aquifer through the recirculation well. The recharge from the recirculation well lowers the interface and as such the effective upconing due to pumping is reduced. However, there are very few reported studies on numerical modeling of flow/transport towards the scavenger and recirculation well systems. A general numerical model for skimming well system is developed that is capable of simulating the flow and salt transport in response to pumping from scavenger and recirculation well systems. The flow and salt transport are assumed to be axis symmetric and having radial as well as vertical components. As such the model is based upon the solution of two coupled differential equations governing two dimensional axis symmetric flow and salt transport in a horizontal confined aquifer. These coupled differential equations are solved numerically invoking the finite difference based Iterative Alternating Direction Implicit Explicit (IADIE) scheme as per the assigned boundary and initial conditions. The model conceptualizes the downstream boundary (i.e. well aquifer interface and the underlying water divide) as a single screen with variable radial well-wards velocity - zero across the blind pipe/water divide, positive across the pumping screen(s) and negative across the recharging screen(s). This strategy renders the proposed model quite well applicable to scavenger and recirculation well systems with closely spaced screens. Constant pressure/concentration condition is assigned across the upstream boundary assumed to be at large enough distance from the well face. No-flow and No-transport boundary conditions are assigned across the lower and upper confining layers. Commencing from assigned initial conditions, the model simulates the pressure and concentration distributions at advancing times. The simulated distributions are further processed to arrive at the drawdown, velocity distribution, production and scavenger well salinities and interface position during pumping and recovery periods. The model is validated invoking an available analytical solution and published field data in respect of partially penetrating and scavenger well systems. Systematic numerical experiments are performed on the proposed model to simulate the response of scavenger and recirculation well systems. The objective is to determine the sensitivity of the production well salinity to the most dominant factors viz. Q2/Q1 (discharge or recharge ratio, where Q] = production discharge and Q2 = scavenging discharge or recirculating recharge) and k:/kr (vertical anisotropy). The numerical experiments reveal that the discharge or recharge through the scavenger or recirculation well effectively reduces the upward velocity component near the interface and the resultant upconing. However, the upconing and hence the production well salinity are found to increase as the vertical anisotropy increases. In the numerical experiments conducted herein, when k^ /kr increases from 0.3 to 1, the increase in the production well salinity is about four times, for an assigned discharge ratio Q2/Q1 = 0.25. The numerical experiments on the recirculation well system reveal that placement of recirculation well screen close to production well greatly reduces the production well salinity provided the recharging water salinity is low. The design of the scavenger well system involves determining the optimal scavenging discharge and positions of production and scavenger wells screens for a given production discharge and hydrogeological parameters. The state variables considered for the design are the production well salinity andthe well face drawdown. A dimensional analysis of the system is carried out to facilitate the design of the scavenger well system in a non-dimensional generalized mode. The well design problem posed herein is viewed as an optimization problem treating dimensionless production and scavenger well screen positions {df, d2d) as the decision variables. The objective is to minimize the dimensionless scavenging discharge {Q2 ) subject to the constraints of restricting the dimensionless steady state production well salinity (Cf) to an acceptable limit and, restricting the dimensionless steady state drawdown (/) to ensure functionality of the production well screen. The optimization algorithm requires computation of the two state variables for several combinations of the two decision variables viz. df and df. To overcome this computation burden, two Artificial Neural Network (ANN) models are developed for the state variables viz. C/ and sd - both at the steady state invoking the numerical model developed herein. The input to theANN models includes the dimensionless production well discharge (Q, ), scavenging discharge (Q2d), screen positions (df, df) and the hydrogeological parameters. These models are computationally quite inexpensive as compared to the numerical model. This facilitates their linkage to an optimizer for determining the positions of the two screens (di, d2) that minimize the scavenging requirement (Q2). The minimization is conducted subject to the constraints that include among others, upper limit onthe production well salinity and the drawdown. To illustrate sensitivity of input parameters on the scavenging requirement, a design example is also presented. Overall results indicate that optimal (minimum) scavenging requirement is quite sensitive to production well discharge, vertical anisotropy and radial intrinsic permeability. It increases as the production well discharge increases or vertical anisotropy increases or radial intrinsic permeability decreases. Further, it is found that the production well screen is optimally located in the upper halfof the aquifer. However, at higher production discharge the production well screen needs to be lowered to accommodate the drawdown. In the numerical experiments performed herein, the production well screen is optimally located at a depth varying from 12% to 26% of the initial freshwater thickness. The optimal in location of the scavenger well screen is found to be near the initial interface position irrespective of the magnitude of the production discharge. IV