numerical analysis of groundwater-flow and solutetransport under skimming well

Upload: zakir-dahri

Post on 05-Apr-2018

217 views

Category:

Documents


0 download

TRANSCRIPT

  • 7/31/2019 Numerical Analysis of Groundwater-Flow and SoluteTransport under Skimming Well

    1/17

    Science, Technology & Development Vol. 30, No. 1 (January-March) 2011

    Science, Technology & DevelopmentVol. 30, No. 1 (January-March) 2011Section B: Biological & Agricultural Research

    Numerical Analysis of Groundwater-Flow and Solute-Transport under Skimming Well

    Zakir Hussain Dahri1, Ata-ur-Rehman Tariq2,

    Bashir Ahmad1, Ghulam Ali1 and Shakil Ahmad3

    The study employed three-dimensional finite-difference groundwater flow (MODFLOW) and

    transport model (MT3D) to model the behavior of groundwater flow and solute transport

    beneath a skimming well. Data regarding aquifer characterization, watertable elevations and

    pumped-water quality for a double-strainer well located at Haroon farm, Khairpur, was used

    to calibrate and verify the flow and transport models. A number of pumping-scenarios wereformulated to evaluate the performance of skimming well.

    The results indicated that, for a shallow layer of fresh water, the pumped-water salinity

    increases linearly with pumping time. The rise in salinity is more pronounced for a higher

    well-discharge and greater well operational factor. Although the drawdown in the well

    achieved steady-state condition within 0.75 day, but the quality of pumped-water continued to

    deteriorate. The rise of saline-water mound also kept rising with pumping time. The study

    revealed that, in the aquifers where salinity-difference of fresh and saline water is small,

    upconing not only occurs at a rapid rate but also to a greater height. Furthermore, it was

    observed that, in case when the saline-water cone had already intruded the fresh aquifer, any

    reduction in well-discharge could neither ensure salt-free water-supply nor any fall in the

    already raised saline-water mound could be observed; instead they kept rising on but at a

    sluggish rate.Unlike the common belief and field observations, intermittent pumping could neither

    control the upward movement of interface, nor was any improvement in quality of pumped

    water observed; instead they remained more or less stagnant, nevertheless the technique is

    helpful to avoid aquifer deterioration. Such contrast in the results of this study is mainly

    attributed to the fact that MODFLOW/MT3D models do not take into account the density of

    fresh and saline water, which is an important parameter in suppressing the upconing.

    Therefore, it was believed that the models tend to over-estimate the solute movement and

    quality of pumped water. Modification is suggested to be incorporated in MODFLOW/MT3D

    programs.

    INTRODU

    ROUNDWATER is a hidden, but impor-

    tant component of the hydrological cycle.

    In many regions, the groundwater

    resource is so huge that its occurrence and

    hydrological significance cannot be overlooked in

    the planning and management of water-resources.

    1

    CTION

    Pakistan Agricultural Research Council, Islamabad, Pakistan.2 Centre of Excellence in Water Resources Engineering, University of Engineering and Technology, Lahore,

    Pakistan.

    12

    3 National University of Science and Technology, Islamabad, Pakistan.

    G

  • 7/31/2019 Numerical Analysis of Groundwater-Flow and SoluteTransport under Skimming Well

    2/17

    Numerical Analysis of Groundwater-Flow and Solute-Transport under Skimming Well

    13

    Groundwater represents an exploitable resource,

    provided its quality and depth permit its

    economical development and optimal use. It can

    serve as a source of dependable, flexible and

    demand-based irrigation-water for boosting

    agricultural production.

    Groundwater has played a major role in

    increasing the overall cropping-intensity inPakistan from about 63% in 1947 to 120% in

    2010. Over 40% of the total irrigation-water at

    farm-gate and over 70% of the total drinking-

    water are currently met from groundwater. The

    spatial distribution of groundwater-abstraction in

    the Country is also highly variable. In some

    regions, annual pumpage has surpassed the safe

    annual yields and so watertables are declining.

    There are however many areas where great

    potential for groundwater-development still

    exists. This requires controlled and optimal

    groundwater-pumpage. Recent estimatesregarding availability and use of fresh

    groundwater indicate that the resource has been

    heavily exploited and is at the brink of

    exhaustion. This large-scale indiscriminate,

    uncontrolled and unregulated abstraction ofgroundwater has changed the policy-approach

    from the development of groundwater to its

    management. The major policy-issues now relate

    to environmental sustainability and long-term

    availability. The standard approaches to managegroundwater often require monitoring of

    groundwater-aquifers, establishment of formalwater-rights and regulation mechanisms, so as to

    bring the groundwater abstraction within

    sustainable levels.

    Sustainable groundwater development in the

    Lower Indus Basin (LIB) of Pakistan is seriously

    hampered by quality-problems, due to intrusion

    of brackish groundwater from adjacent areas,

    upconing of underlying saline water, backward

    intrusion of seawater in coastal areas and disposal

    of drainage effluents. The native groundwater

    was originally brackish, because of the marine

    origin of the underlying geologic formation. Thisnative brackish groundwater is now overlain by a

    fresh-water layer of varying thicknesses, due to

    seepage from conveyance system and percolation

    from irrigation and precipitation. The depth of

    that fresh-water layer is greater near the

    recharging sources and decreases away from the

    line source. Subsurface investigations in the LIB

    show that, for about 4.66 m. ha (75.77 per cent)

    of the irrigated area the underlain groundwater is

    classified as hazardous and saline (TDS > 3,000

    ppm), while for another 8.94 per cent of the area,

    the groundwater is marginal (TDS from 1,500 to

    3,000 ppm) and for only 15.29 per cent of the

    area is the groundwater fresh (TDS < 1,500 ppm)(Zuberi and Sufi, 1992). The areas with fresh

    groundwater thickness less than 45 m can be

    termed as critical areas, where sustainable

    groundwater development requires careful

    thinking in the selection, design and operation of

    irrigation-wells. Moreover, the spatial distribution

    of its abstraction is highly variable. In some

    regions, annual pumpage has surpassed the safe

    annual yields and the watertables are declining.

    There are however many areas where great

    potential for groundwater-development still

    exists. This requires controlled and optimalpumpage of groundwater.

    Under natural conditions, the fresh and saline

    groundwater layers are usually in a state of

    dynamic equilibrium. Extraction from fresh-water

    layer, however, disturbs that equilibrium and,

    consequently, the adjacent or underlying saline

    groundwater starts moving upward in the shape of

    a cone. This phenomenon is termed as saline-

    water upconing. The extent and intensity of

    upconing is greatly dependent on pumping-rates,

    degree of penetration, thickness of fresh

    groundwater layer, quality of water-withdrawalzone, location of the fresh and saline water

    interface, vertical hydraulic conductivity, and

    extent of aquifer-recharge. Moreover, hydro-

    dynamic-dispersion phenomenon significantly

    affects the solute transport towards the well

    (Kemper, et al., 1976; Mirbahar, et al., 1997 and

    Sufi, et al., 1998). Therefore, many engineers and

    researchers are of the view that the overlying

    fresh groundwater must be abstracted in such a

    way that the underlying saline groundwater is not

    disturbed (Chandio and Chandio, 1992, Awan

    1990). Properly designed and operated skimming-wells may offer an economical and sound

    alternative over conventional deep wells, where

    native saline groundwater exists at about 30 m

    below ground-surface (Chandio and Chandio,

    1992). However, very few researchers in Pakistan

    have studied the upconing-phenomena beneath a

  • 7/31/2019 Numerical Analysis of Groundwater-Flow and SoluteTransport under Skimming Well

    3/17

    Science, Technology & Development Vol. 30, No. 1 (January-March) 2011

    skimming well, and only limited information is

    available in this regard (Sufi, 1999). The present

    research study is aimed at analyzing the

    groundwater-flow and solute-transport

    phenomena under a skimming well, and

    evaluating the impact of operational strategies on

    quality of pumped water and upconing of

    underlying brackish water. The research outcomewill be helpful to formulate guidelines for

    sustainable development of fresh groundwater in

    the areas where a shallow layer of fresh water

    overlies the native brackish groundwater.

    THEORETICAL CONSIDERATIONS

    The theoretical considerations regarding

    fresh-water skimming can be divided into two

    categories i.e. Immiscible and Miscible Theory.

    The first formulation is less rigorous and less

    realistic, assuming the application of DF

    assumptions, and existence of sharp interfacehaving abrupt transition from a relatively fresh to

    highly saline water. The second formulation is

    more realistic but, at same time, more

    complicated assuming change in salinity from salt

    water to fresh water under the hydrodynamic

    dispersion, and existence of the transition-zone of

    relatively large thickness between the fresh and

    salt water. The transition zone is characterized by

    a gradual change in the concentration of water

    from that of fresh water to lower saline water.

    Both these formulations are, however, important

    and are discussed in some detail in the followingsections.

    Physics of groundwater flow and solute-

    transport mechanisms

    Groundwater flow is described by Darcys

    law, which states that flow-rate is proportional to

    the hydraulic gradient, where the constant of

    proportionality is hydraulic conductivity

    describing the hydraulic properties of the

    medium. The hydraulic conductivity varies in

    space in a manner that, to large degree, is tied up

    with the spatial variation of the geologicalproperties. Darcys law is a well-proven

    relationship, which has frequently been used for

    quantitative assessments of groundwater flow.

    Mathematically, it can be formulated in a tensor

    notation as:

    t

    hSq

    x

    hK

    xs

    j

    j

    i

    =+

    ; i,j = 1,2,3(1)

    where q is specific discharge (LT-1

    ), Kij is

    hydraulic conductivity (LT-1

    ), h is hydraulic head

    (L), xi is space coordinate (L).

    Combination of Darcys flow-equation withcontinuity-equation, which represents the

    conservation of fluid mass, yields the following

    partial differential equation describing the three-

    dimensional movement of groundwater through

    porous media:

    szyx qz

    hK

    zy

    hK

    yx

    hK

    x+

    +

    +

    t

    hS s

    =

    ....(2)

    where Kx , Ky , Kz are the values of hydraulic

    conductivity along x, y, z coordinates, which are

    assumed to be parallel to the major axes of

    hydraulic conductivity (LT-1); qs is the fluid

    sink/source term or volumetric rate, at which

    water is added or removed from the system per

    unit volume of aquifer; Ss is the specific storage

    or volume of water released from storage in a unit

    volume of aquifer per unit decline in head.

    While flow of groundwater is governed by

    Darcys law, the transport of solute in a

    groundwater-system is controlled by manyfactors. The most important mechanisms affecting

    the transport of solute in a porous medium are:

    advection or convection, hydrodynamic

    dispersion, and various chemical reactions and

    decay phenomena, which may be regarded as

    source-sink phenomena for the solute. All these

    phenomena cause changes in the concentration of

    solute in the flowing fluid. In general, variations

    in solute-concentration cause changes in liquids

    density and viscosity. These, in turn, affect the

    flow regime (i.e. velocity-distribution) that

    depends on these properties. The partialdifferential equation describing the three-

    dimensional transport of contaminants in

    groundwater (Javendal et al., 1984) is written as

    follows:

    14

  • 7/31/2019 Numerical Analysis of Groundwater-Flow and SoluteTransport under Skimming Well

    4/17

    Numerical Analysis of Groundwater-Flow and Solute-Transport under Skimming Well

    ( ) =

    ++

    = N

    k

    ks

    s

    i

    ii

    ij

    i

    RCq

    Cvxx

    CD

    xt

    C

    1

    (3)

    where,

    C = concentration of contaminants dissolved in groundwater, M/L3

    t = time, Txi = distance along respective cartesian-coordinate axis, L

    Dij = hydrodynamic dispersion coefficient, L2/T

    vi = seepage or linear pore-velocity, L/T

    qs = volumetric water-flux per unit aquifer volume sources (+) & sinks (-), T-1

    Cs = concentration of the sources or sinks, M/L3

    = porosity of the porous medium, dimensionlessRk = chemical reaction sink/source term, representing the rate of change in solute mass of a

    particular species, due to N chemical reactions.

    METHODOLOGY

    Data collection: The skimming well selected forthis study is located at Haroon Farm near Gambat

    railway station, district Khairpur, and is part of

    Khairpur SCARP area. During the past few years,

    the area has been plagued by waterlogging and

    salinity problems as the watertable generally

    fluctuates between 0.5 and 1.0 m. The occurrence

    of highly saline groundwater at depths ranging

    from 25 to 35 m has restricted the installation of

    large-capacity tubewells. The skimming-well

    consists of two strainers, each bored at 12.2 m

    apart and discharging 0.4265 cfs. Specification

    and borehole lithology, along with profilicdistribution of salinity, is described Tables 1 and

    2, respectively. The penetration of both the bores

    was 19.82 m, with 10.67 m strainer surrounded

    by coconut fiber. The internal diameter of the

    blind pipe and strainers are 10.2 and 20.4 cm,

    respectively. Both the bores are connected withcentrifugal pump, driven by 15 HP diesel engine.

    The data-requirements of PMWIN are

    categorized into time-constant and time-variant

    data. Time-constant data include aquifer

    geometry (areal and vertical distribution of

    subsurface strata, aquifer-thickness, etc) and

    hydraulic parameters (hydraulic conductivities,

    effective porosity, specific yield, specific storage,

    etc). Time-variant data are recharge, pumping,

    evapotranspiration, groundwater levels, ground-

    water quality, etc. Time-constant data was taken

    from previous studies, such as Lower IndusReports, ACE 1997, ACE 2001, while time-

    variant data for this skimming well was collected

    by Drainage and Reclamation Institute of

    Pakistan, Tando Jam.

    Table-1: Specification of piezometers and boreholes dimensions.

    Piezometer

    No.Distance From the

    Centre of Well (m)Diameter

    (cm)Blind Pipe

    (m)Strainer

    (m)Total Length

    (m)

    PZ1 9 3.18 3.04 1.52 4.56

    PZ2 26 3.18 3.04 1.52 4.56PZ3 77 3.18 3.04 1.52 4.56

    PZ4 168 3.18 3.04 1.52 4.56

    Bore-1 0 10.2 9.15 10.67 19.82

    Bore-2 0 10.2 9.15 10.67 19.82

    15

  • 7/31/2019 Numerical Analysis of Groundwater-Flow and SoluteTransport under Skimming Well

    5/17

    Science, Technology & Development Vol. 30, No. 1 (January-March) 2011

    16

    Table-2: Soil lithology and salinity distribution in both the boreholes of the skimming well at

    Haroon farm, Khairpur.

    Bore-1 Bore-2Depth (m)

    Soil Texture Salinity (ppm) Soil Texture Salinity (ppm)

    0 3 silty loam 896 silty clay loam 768

    3 6 very fine sand 960 fine sand 838

    6 9 medium sand 1088 very fine sand 960

    9 12 coarse sand 1152 very fine sand 1280

    12 15 coarse sand 1024 Medium sand 1152

    15 18 coarse sand 1152 coarse sand 1152

    18 21 coarse sand 1024 coarse sand 1216

    21 24 - - - 1152

    24 27 - - - 1568

    27 30 - - - 3410

    60 - - - 8064

    Model formulation: Numerical modeling is

    commonly used for simulating the complex

    groundwater systems, possessing non-linear

    features, that may not be solved through

    analytical approaches. It permits prediction of the

    response of an aquifer to applied stresses and

    evaluates alternative suggestions for its use. Of

    the number of groundwater-models available, the

    PMWIN (Chiang and Kinzelbach, 1996), which isa complete simulation system for modeling

    groundwater-flow, with MODFLOW (Harbaugh

    and McDonald, 1988, 1996), and solute-transport

    processes with MT3D (Zheng, 1996), was

    selected for this study. Using these models,

    different scenarios can be developed and run to

    study fluctuations of water-table and upconing

    phenomenon in fresh and saline groundwater

    systems.

    The important steps for model-formulation

    are: spatial and vertical discretization of the grid-

    system; setting up initial and boundary

    conditions; temporal discretization; specifying

    aquifer-parameters; and assigning external stress

    packages, like wells, rivers, recharges, evapo-

    transpiration, advection, dispersion, etc.

    The mesh was generated in a non-uniform

    manner, the mesh spacing being smaller near the

    well and larger away from the well at a constant

    ratio of 1.4. Moreover, the grid spacing was kept

    small enough to avoid any artificial oscillation

    and numerical dispersion. The grid size varied

    from 3.05 m to 11.72 m. The total grid domain

    (417.55 m 417.55 m) was divided into 59columns and 59 rows. The generated model grid-

    system is shown in Figure 1, while Table-3

    describes spatial discretization of the model grid.

    The aquifer was discretized into 8 layers

    (Table-4) on the basis of soil-texture, salinity-

    distribution, length of blind pipe and well-screen,

    expected drawdown and ease in simulation, to

    achieve the planned objectives.

    The average water-table elevation within the

    model domain, prior to pumping was 119.136 m.

    This head was specified as the initial hydraulic

    heads, throughout the domain. The spatial and

    vertical distribution of initial concentration was

    estimated from the bore log of the well and

    piezometers installed around it. The heads and

    concentrations specified as the starting values, at

    the boundaries, were set to be constant, i.e. did

    not change with time, throughout the simulation

    period. A no-flow boundary was used at the base

    of the system, representing the assumed zero

    movement of water into or out of the relatively

  • 7/31/2019 Numerical Analysis of Groundwater-Flow and SoluteTransport under Skimming Well

    6/17

    Numerical Analysis of Groundwater-Flow and Solute-Transport under Skimming Well

    Constant head and constant concentration boundary

    17

    Figure 1. Spatial discretization of the grid system

    Table-3: Spatial discretization of the grid system

    S. No. No. of Columns No. of Rows Grid Size (m x m)

    1 7 7 3.05 x 3.05

    2 10 10 4.27 x 4.27

    3 16 16 5.98 x 5.98

    4 14 14 8.37 x 8.37

    5 12 12 11.72 x 11.72

    Total 59 59 417.55 x 417.55

    Table-4: Vertical discretization of model domain

    LayerNo.

    Layer

    Thickness

    (m)

    Cumulative

    Thickness (m)Top

    Elevation

    Layer (m)

    Bottom

    Elevation

    Layer (m)

    Concentration

    (ppm)

    1 9.15 9.15 120 110.85 1000

    2 5.335 14.485 110.85 105.515 1100

    3 5.335 19.82 105.515 100.18 12304 8.18 28 100.18 92 1400

    5 5 33 92 87 2000

    6 7 40 87 80 3500

    7 20 60 80 60 7000

    8 60 120 60 0 10000

  • 7/31/2019 Numerical Analysis of Groundwater-Flow and SoluteTransport under Skimming Well

    7/17

    Science, Technology & Development Vol. 30, No. 1 (January-March) 2011

    18

    impermeable bedrock. In case of transport-model,

    a constant concentration-boundary was assigned

    at the lowest layer. The skimming-well was

    hydraulically simulated as a well pumping from

    an infinite aquifer, where head and solute

    concentration in the cells beyond the radius of

    influence will remain constant and equal to the

    initially determined values.The temporal discretization include the time-

    unit and the number of stress-periods, time steps

    and transport steps in each time step or stress

    period. In MODFLOW, the simulation-time can

    be divided into stress periods i.e., the time-

    intervals during which all external excitations or

    stresses are constant which, in turn, may be

    divided into time-steps. In MT3D model, each

    time step is further divided into smaller time-

    increments, called transport-steps, because the

    length of a time step used for a head solution may

    be too large for a transport-solution. The study

    used different scenarios and simulation-time,

    number of stress-periods; and the number of time-

    steps in each stress-period varied for each

    scenario. The length of transport-step in each

    simulation, however, was set to be calculated

    automatically by the model itself.Site-specific data, regarding aquifer

    parameters, is not available. However, as per

    elaborated pumping tests performed around the

    study area during the regional groundwater

    investigations (during the Lower Indus Project),

    some estimates are available. These were refined

    during model calibration and verification. The

    final values used for this study are summarized in

    Table-5.

    Table-5: Parameter values used for simulation

    S. No. Parameter Value

    1 Initial Watertable Elevation 119.136 m.

    2 Initial Concentration As per bore log (Table 4)

    3 Horizontal Hydraulic Conductivity(Kh ) 20.16 m/d

    4 Vertical Hydraulic Conductivity (Kv ) 6.72 m/d

    5 Anisotropy Ratio (Kh / Kv ) 3

    6 Effective Porosity () 13 %7 Specific Yield (Sy) 7 %

    8 Specific Storage (Ss) 110-5

    9 Recharge 361 mm/year

    10 Concentration of Recharge Flux 550 ppm

    11 Advective parameters for particle placement/movement

    As per users manual

    12 Longitudinal Dispersivity 8 m

    13 Ratio of horizontal transverse dispersivity tolongitudinal dispersivity

    0.33

    14 Ratio of vertical transverse dispersivity to

    longitudinal dispersivity

    0.1

    15 Effective molecular diffusion coefficient 0.00139

  • 7/31/2019 Numerical Analysis of Groundwater-Flow and SoluteTransport under Skimming Well

    8/17

    Numerical Analysis of Groundwater-Flow and Solute-Transport under Skimming Well

    19

    There are two well-points acting as double-

    bore skimming-well for this particular study. The

    bores are located 12.2 m apart. The length of

    blind pipe for both the bores is 9.15 m, which is

    represented by top model layer (layer-1). The

    length of screen for both the well-points is 10.67

    m, which is divided into two model layers

    (layers-2 and 3). The designed discharge for eachbore-point is 43.5 m3/hr (0.4265 cfs).

    MODFLOW assumes uniform distribution of

    water entering into the well-screen throughout its

    length, which is not realistic as, in actual field

    conditions, more water enters into the well-screen

    from the top and gradually decreases towards the

    bottom. Li (1954) and Soliman (1965) analyzed

    the flow to a well and showed that the entrance-

    velocity in the upper 10 % of the well-screen was

    70 times that in the lower 10% in an ideal aquifer.

    If the open area of the screen is constant

    throughout, we can approximate the distributionof flow rate into the screen. Based on Li and

    Solimans statement, it was found that about 77%

    of the water enters from the upper half of the

    well-screen (layer-2) and the remaining 23% from

    the lower half of the screen (layer-3). Thisdistribution of water-entry to the screen was used

    to determine the quality of pumped water.

    Model calibration and verification: Calibration

    is an inverse modeling process, in which certain

    model input-parameters are adjusted until model-

    output parameters/dependent variables match the

    field-observed values to an acceptable degree.The data regarding watertable elevations,

    pumped-water salinity, and profilic distribution of

    groundwater-quality were used to calibrate both

    the flow (MODFLOW) and transport (MT3D)

    models. Simple manual trial-and-error method

    was used to sequentially calibrate both the

    models, against the data observed after 15.67 hrs.

    of continuous well-operation, because at this

    stage the well-drawdown tends to achieve steady-state condition.

    Model verification is a process in which the

    calibrated model is shown to be capable of

    reproducing a set of field-observations

    independent of that used in model calibration.

    The calibrated models were verified against the

    observed data at different pumping-periods

    (10.92, 13.67 and 17.67 hrs.) through an

    interactive procedure of sequential calibration and

    recalibration, until the models were capable of

    reproducing all the data-sets used for model

    verification, including the one used for modelcalibration. Besides graphical comparison of

    observed and simulated values, some statistical

    measures of goodness-of-fit were also used for

    qualititative comparison and assessment of the

    model calibration. Figures 2 (c-f) and 3 presentcomparison of observed and simulated hydraulic

    heads and pumped-water salinities at different

    pumping periods, while Table 6 shows statistical

    analysis of model calibration and verification.

    There is good agreement between the observedand simulated values, and so the models are

    capable of reproducing the field observations atany specified time.

    Table-6: Statistical analysis of model calibration and verification.

    Simulation

    Time(hrs)

    Liner Correlation

    Coefficient(r)

    Root Mean Squared

    Residual Errors (RMS)

    Maximum

    Error

    (ME)

    Modeling

    Efficiency

    (EF)

    MODFLOW for Hydraulic Heads

    10.92 0.522064534 0.042077524 7.92E-08 0.98307

    13.67 0.5322365 0.033377298 9.04E-11 0.98956115.67 0.522615471 0.044022472 1.63E-07 0.981527

    17.67 0.513007667 0.055299168 2.09E-06 0.970306

    MT3D for Pumped Water Salinity

    1.0 - 17.67 0.99313918 6.091711582 4.96 0.962462

  • 7/31/2019 Numerical Analysis of Groundwater-Flow and SoluteTransport under Skimming Well

    9/17

    Science, Technology & Development Vol. 30, No. 1 (January-March) 2011

    Calibration (15.67 hrs)

    118

    118.2

    118.4

    118.6

    118.8

    119

    119.2

    119.4

    0 50 100 150 200

    Distance from centre of wells (m)

    Piezome

    trichead(m)

    Initial

    Obseved

    Simulated

    Verification (10.92 hrs)

    118

    118.2

    118.4

    118.6

    118.8

    119

    119.2

    119.4

    0 50 100 150 200

    Distance from centre of wells (m)

    Piezom

    etrichead(m)

    Observed

    Simulated

    c. Hydraulic heads after 15.67 hrs of operation d. Hydraulic heads after 10.92 hrs of operationVerification (13.67 hrs)

    118

    118.2

    118.4

    118.6

    118.8

    119

    119.2

    119.4

    0 50 100 150 200

    Distance from Well (m)

    PiezometricHead(m

    .

    Observed

    Simulated

    Verification (17.67 hrs)

    118

    118.2

    118.4

    118.6

    118.8

    119

    119.2

    119.4

    0 50 100 150 200

    Distance from centre of wells (m)

    Piezometriche

    ad(m)

    Observed

    Simulated

    e. Hydraulic heads after 10.92 hrs of operation f. Hydraulic heads after 17.67 hrs of operationFigure 2 Observed and simulated hydraulic heads at various pumping times.

    Calibration and verification

    1100

    1130

    1160

    1190

    1220

    1250

    1280

    0 2 4 6 8 10 12 14 16 18 20

    Pumping time (hrs.)

    Pumpedwatersalinity(ppm)

    Observed

    Simulated

    Figure 3. Observed and simulated pumped water salinities at various pumping times

    20

  • 7/31/2019 Numerical Analysis of Groundwater-Flow and SoluteTransport under Skimming Well

    10/17

    Numerical Analysis of Groundwater-Flow and Solute-Transport under Skimming Well

    RESULTS AND DISCUSSION

    It is imperative to maintain the original

    dynamic equilibrium between the layers of fresh

    and brackish water, if groundwater is to be

    developed on sustainable basis. Thus, there is an

    important need to investigate the mechanics of

    solute-transport in response of fresh groundwater

    withdrawal. For this purpose, an effort has beenmade to replicate the actual field-conditions of

    pumpage by the numerical model, so that one

    may have a close look at the flow and solute-

    transport phenomena under a skimming well.

    Model simulations were performed for three

    different scenarios.

    The 1st scenario represents an ideal situation

    where the well is operated intermittently and

    exact time of its operation and closure is

    specified. The total simulation-time is divided

    into 60 stress-periods, each representing the timein which the well was either on or off. The main

    problem associated with this type of scenario is

    small simulation period, as PMWIN can simulate

    a maximum of 80 stress-periods while in actual

    field-conditions the majority of wells are operated

    on daily basis (operated for some time and closed

    for rest of the day). In this way, the maximum

    simulation-time can be only 40 days. Anyway, to

    get an idea of the situation, simulation time of

    scenario-1 was kept 30 days, having 60 stress-

    periods.

    The simulation time of scenario-2 was alsokept 30 days, but the entire time was considered

    single-stress period, in which the well was

    operated continuously at a reduced discharge i.e.

    designed well-discharge was multiplied with the

    well-operational factor. The total volume of

    pumped water for both the scenarios, however, is

    the same. Although, this scenario is not realistic,

    but it does provide an opportunity to simulate the

    models for a desirable time-period. This scenario

    was designed just to compare its outcome with

    the previous one. Figure 4 depicts the pumping

    schedule and well discharge for both thescenarios, while pumped-water quality is

    presented in Figure 5.

    0

    400

    800

    1200

    1600

    2000

    0 5 10 15 20 25 30

    Pumping Duration (days)

    WellDischarge(m3/day)

    Scenario-1 Scenario-2

    Figure 4. Graphical representation of well discharge for scenarios - 1 & 2

    21

  • 7/31/2019 Numerical Analysis of Groundwater-Flow and SoluteTransport under Skimming Well

    11/17

    Science, Technology & Development Vol. 30, No. 1 (January-March) 2011

    1100

    1200

    1300

    1400

    1500

    0 2 4 6 8 10 12 14 16 18 20 22 24 26 28 30

    Pumping Duration (days)

    PumpedWate

    rQuality

    (ppm)

    Scenario-1 Scenario-2

    Figure 5. Graphical representation of pumped-water quality for scenarios - 1 & 2

    The pumped water quality for both the

    scenarios clearly indicates insignificant difference

    between their outcomes. This insignificant

    difference encouraged us to design and adopt 3rd

    scenario, by combining the 1st and 2nd scenarios.

    This scenario can employ longer simulation

    period and very closely matches the actual field-

    conditions, where temporal variation of pumping

    is quite high, which is mainly dependent on

    availability of canal water, crop water-

    requirements and climatic conditions. The total

    simulation time of 565 days is divided into 10

    stress-periods. An increased well-operation factor

    is assigned for periods during which canal water

    supplies are very low and the demand is high. The

    well is kept closed for periods during which either

    there was no crop in the field or, because of

    heavy rainfall in monsoon season, crops do not

    require supplemental irrigation. The pumping

    schedule, well-discharge and quality of pumped

    water for 3rd scenario are presented in the

    Figure 6, whereas temporal variability in the

    aquifer salinity at different observation-depths is

    presented in Figure 7.

    0

    100

    200

    300

    400

    500

    600

    0 100 200 300 400 500 600

    Pumping Duration (days)

    W

    ellDischarge(m3/day)

    0

    500

    1000

    1500

    2000

    2500

    3000

    Pum

    pedWaterQuality(ppm)

    Well Discharge Pumped Water Quality

    Figure 6. Temporal variation of well discharge and pumped-water quality for scenario-3

    22

  • 7/31/2019 Numerical Analysis of Groundwater-Flow and SoluteTransport under Skimming Well

    12/17

    Numerical Analysis of Groundwater-Flow and Solute-Transport under Skimming Well

    500

    1500

    2500

    3500

    4500

    5500

    6500

    0 100 200 300 400 500 600

    Pumping Duration (days)

    Salinityrise(ppm)

    4.575 m

    11.82 m17.15 m

    23.91 m

    30.5 m

    Obsevation

    depth from

    ground

    surface

    Figure 7. Temporal variation of aquifer salinity at different observation depths

    Figures 5, 6, and 7 further indicate that, for

    shallow fresh-water layer, the salinity rise is alinear function of operational time. The rise in

    salinity is more pronounced for higher well-

    discharge or greater well-operational factor. The

    results also indicated that the design of the

    selected skimming-well is faulty for the given

    aquifer properties and profilic distribution of the

    salinity within the aquifer. Zuberi and McWhorter

    (1973) suggested that, for aquifer properties in

    Pakistan, the discharge of individual skimming

    well should be in the range of 0.1 to 0.3 cfs,

    whereas the discharge of the selected skimming

    wells was 0.853 cfs, which seems too high.Moreover, there is very little available fresh-

    water cushion below the well-screen. The

    optimum well-penetration ratio of the site under

    study was found to be 23 %, against the present

    penetration of 33%.

    It can also be observed from Figure 7 that,despite the fact that penetration of well-screen

    started from the depth of 9.15 m to 19.82 m from

    the ground surface, the salinity of the upper fresh

    layer above 9.15 m depth was increased, whereas,

    it is quite clear that flow in that portion is radially

    downward. Moreover, surface recharge alsooccurs in that layer, so the question is what

    prompted the salinity to increase there. Salinity

    rise in the top layer above well-screen may be

    attributed to the process of molecular diffusion

    and transverse dispersivity, which generally

    propagate in the direction orthogonal to the flow.

    Effect of drawdown: The drawdown for

    different discharges is shown in Figure 8, whichindicated that, for each well discharge, there

    existed a limit on drawdown at which the

    discharge and recharge balanced each other and

    ultimately steady-state condition was achieved.

    Figure 9 shows the spatial extent of upconing at

    various observation depths after 335 days of

    simulation. It can be observed from this figure

    that the radius of influence of the saline-water

    mound at all observation depths is about 150 m,

    which is roughly half of the radius of influence of

    the cone of depression. Therefore, it can be

    approximated that if the well is to be replaced byanother well, due to salinity rise under the well,

    then the new well must be 300 to 500 m away

    from the previous one.

    But, unlike the steady-state condition of

    drawdown, the quality of pumped water

    continued to deteriorate and also the rise of salt-water mound, which kept rising on with pumping

    time. It is commonly believed that smaller

    upcoming occurs for smaller drawdowns and that,

    to limit the upconing, it is essential to limit the

    drawdown. Nevertheless, it is not directly

    proportional to the drawdown. In fact, drawdownitself is a dependent variable and is the function

    of well-discharge and certain aquifer properties,

    and can be limited by decreasing the well-

    discharge; whereas upconing is related with many

    factors, including those which affect drawdown.

    Therefore, relating it with drawdown or any other

    factor alone is not correct.

    23

  • 7/31/2019 Numerical Analysis of Groundwater-Flow and SoluteTransport under Skimming Well

    13/17

    Science, Technology & Development Vol. 30, No. 1 (January-March) 2011

    118.90

    118.95

    119.00

    119.05

    119.10

    119.15

    119.20

    -300 -200 -100 0 100 200 300

    Distance from Well (m)

    Watertab

    leElevation

    (m)

    well Q = 522 m3/d

    well Q =348.7 m3/d

    well Q = 261 m3/dWell is closed

    Figure 8. Cone of depression for different well-discharges after 335 days of simulation

    After 335 days of simulation

    0

    2500

    5000

    7500

    10000

    12500

    15000

    -300 -200 -100 0 100 200 300

    Distance from the centre of wells (m)

    EC(ppm)

    118.0

    118.2

    118.4

    118.6

    118.8

    119.0

    119.2

    Hydraulichead(m)

    4.575 m

    11.82 m

    17.15 m

    23.91 m

    30.5 m

    Observation

    depth from

    ground

    surfacecone of dipressionupconing

    Figure 9. Rise of salt water mound at various observation-depths, after 335 days of pumping

    Bower (1978) declared that, if the

    groundwater withdrawal from the coastal aquifer

    exceeds the safe yield and the water-levels

    decline, the salt-water would rise 40 m for everymeter of drop in the watertable. However, during

    this pumping-scenario, the maximum drawdown

    only 0.22 m occurred against the discharge of 522m3/d, but the fresh and salt water interface was

    raised by 19.18 m in 335 days of well operation,

    which is 87 times greater than the drawdown.

    The starting salinity-level of the fresh and saline

    interface is considered to be 3,500 ppm. The large

    difference between Bowers statement and the

    simulated results of this scenario can be attributed

    to the fact that his statement is primarily based onthe Ghyben-Herzberg relation, which does not

    take into account the vertical component of

    groundwater-flow; hence, it underestimates theextent of upconing. Furthermore, the small

    difference between the salinities of fresh and

    saline waters of the aquifer under study, viz-a-viz

    seawater, also causes the saline water to move

    24

  • 7/31/2019 Numerical Analysis of Groundwater-Flow and SoluteTransport under Skimming Well

    14/17

    Numerical Analysis of Groundwater-Flow and Solute-Transport under Skimming Well

    25

    more quickly. It reveals that in the aquifers where

    salinity-difference of fresh and saline waters is

    small, upconing not only occurs at a rapid rate,

    but also to a greater height.

    Effect of discharge reduction: Obviously, the

    higher discharges induce greater upconing.

    Therefore, if the well discharge is low, right from

    the beginning of pumping, then the rise ofinterface will be slower. What of the situations

    where rise of interface has already taken place

    due to higher well-discharge: can any reduction

    in well-discharge suppress the upcoming? This

    question is answered in the following paragraphs.

    At the initial stage of pumping, much of thewater contribution is from top of the screen. As

    the pumping continues, the area and depth of the

    propagation of streamlines are increased. At the

    interface, the streamlines tend to flow tangentially

    to the top of interface. All other streamlines,

    which originate below these ones, will bring salt

    water to the well, thereby deteriorating the quality

    of pumped water and cause rising of salt-water

    mound. Now if the salt water has already intruded

    into the fresh-water zone, reduction in well

    discharge will neither ensure salt-free water

    supply nor can any fall in the raised mound of salt

    water be observed. Figs. 7 and 8 confirm such a

    statement.

    These figures showed that, even when the

    well-discharge was halved from 522 m3/d to 261

    m

    3

    /d, no improvement in pumped water qualityand recession of salt-water cone could be

    observed; instead they kept on rising. This is

    because, even though the discharge was reduced,

    the flow towards the well continued through the

    same streamlines but at a sluggish rate. The

    streamlines, which originated from salt-water

    zone, will vanish eventually only when the well is

    stopped and groundwater is fully recovered, but

    the position of interface would be higher than the

    initial one. Therefore, when the well is started

    next time, it might have less available thickness

    of fresh water below the well-screen and chances

    of upconing will be greater.

    Effect of intermittent pumping: Generally

    speaking, when a partially penetrating well

    continuously discharges water from an

    unconfined aquifer, in which fresh water is

    underlain by salt water and is separated by a well-

    defined interface located within the reach of

    streamlines (generated due to pumping of well

    from the fresh water zone), then ultimately a time

    must come when the salt-water of interface will

    enter the well-screen, causing deterioration of

    pumped water quality. It is commonly believed

    and also has been witnessed during the field

    studies (Saeed, et al., 2002 and Ashraf, et al.,

    2001) that intermittent pumping may control theupward movement of fresh and salt-water

    interface. The well discharge and operational

    factor, however, must be low enough, not

    allowing the streamlines to reach the interface.

    During the periods of well-closure, the generated

    streamlines will tend to stabilize but the time

    required for their complete stabilization may be

    quite long. At complete stabilization of

    streamlines, the well may be turned on again for a

    certain safe period, at which regenerated

    streamlines are not allowed to reach the interface.

    This safe period of pumping depends on well-discharge and operational factor, aquifer

    characteristics, fluid properties and fresh-water

    cushion below the well-screen. There is need to

    determine this safe pumping period for a given set

    of conditions, but this might be a very difficult

    task, and some-times not practicable. McWhorter

    (1980) recommended a minimum of 5 times the

    length of pumping-period as the rest period for

    the well between the two pumping periods.

    The well under study was operated

    intermittently at varying operation-factors and

    was closed for sufficiently long time, during andafter the pumping, to observe the rates of rise and

    recession of the salt-water cone. The quality of

    pumped water deteriorated linearly as the

    pumping continued. During the periods of well-

    closure, the cone remained more or less stagnant.

    Again from the Figures 4.8 and 4.9, one can

    easily judge that the impact of well-closure on

    well-water quality and fall of salt-water cone is

    negligible, which is totally in contrast with the

    actual field conditions. This is well supported by

    the fact that, like many other wells in the Region,

    this well too discharges an acceptable quality ofgroundwater for some time and is closed for a

    sufficient time to allow the fresh groundwater to

    regain its original position, through surface as

    well as lateral recharge. Moreover, Hafeez et al.,

    1986 conducted field studies at Mona project,

    Bhalwal, to determine the rates of rise and fall of

  • 7/31/2019 Numerical Analysis of Groundwater-Flow and SoluteTransport under Skimming Well

    15/17

    Science, Technology & Development Vol. 30, No. 1 (January-March) 2011

    26

    the salt-water cone beneath a skimming well.

    They witnessed the rise of cone by 30 feet in 15

    days of continuous operation at 0.5 cfs discharge,

    whereas, the recession of cone was much slower

    than its rise: it took 164 days to fall only two-

    third (20 ft) of rise. They also showed that about

    half of the rise could be made to recede by

    stopping the well for a period equal to twice theoperation time.

    The negligible effect of well-closure on fall

    of the salt-water mound, indicated by the model

    results, is clearly because the MT3D model does

    not consider the density of water, which creates a

    downward potential during well-closure, thereby

    forcing the cone to recede. The effect of density

    will be large if the salinity difference between

    fresh and saline water is greater. It is therefore

    believed that the MT3D model might

    overestimate rise of cone and can not properly

    simulate the intermittent pumping of well.Nevertheless, the model is a valuable tool for

    simulating the sources or sinks, which are usually

    continuous in nature.

    Suggested modification: The results of the

    previous scenarios led to the conclusion that the

    MT3D model could not accurately simulate the

    intermittent pumping, whereas its role in

    abstraction of fresh groundwater, while keeping

    the rising of underlying salt-water well under

    control, can not be overlooked. . Therefore, it was

    strongly felt that there is need to introduce some

    modification so that the model is able to simulatemore accurately the well that is not continuously

    operating. The idea was primary based on the

    theory described by Sehni (1973) that the two

    fluids (fresh and saline water) are miscible and in

    reality, at their contact, they tend to mix with

    each other by molecular diffusion and

    macroscopic dispersion. Thus, they are not

    separated by an oil-water type interface; they do

    not constitute distinct fluid phases, and there is no

    pressure discontinuity where they are

    encountered. Since it is assumed that no pressure

    discontinuity exist across the interface, thepressure at any point within the aquifer can be

    taken as constant:

    P1 = P2 = P3 = = Pn

    1gh1 = 2gh2 = 3gh3 = . = nghn

    The model domain of this study was divided

    into eight layers, each having different salinity.

    The above equation can be reduced as,

    1h1=2h2 = -------------------- = 8h8

    If h1 is considered as the watertable elevation

    in the top layer, which is known, then the

    hydraulic heads in the lower layers can beapproximated by putting the value of density of

    water in each layer in the above equation.

    Rubin and Pinder (1977) assumed a linear

    relationship between salinity and specific weight

    as follows;

    = f(1 + c)

    where, c is salinity, and is coefficient relatingchanges in density to concentration.

    Sufi (1999) determined the densities of saline

    waters of varying concentrations. From these

    known values of densities, the density of fresh

    water and were calculated as 1.00166845gm/cm3 and 0.000000785 respectively. Putting

    these values in the above equation, one can

    calculate the density of any specified

    concentration. The corrected head (h*) becomes

    as h* = h/, where h is the head computed byMODFLOW.

    The MODFLOW and MT3D programs are

    operated in sequence. The MODFLOW generates

    head-distribution for all cells for all time-steps.

    The MT3D program subsequently uses thesehead-distributions to generate time-variant flow-

    velocity field and then determines the solute

    movement, due to advection and dispersion. It is

    required that MODFLOW and MT3D should be

    made to execute recursively over small time-

    increments. At each time-increment (including

    initial condition), the hydraulic head potential be

    adjusted according to the salt-contents in each

    cell, as per above equation. This will cause the

    hydraulic gradient of cells in lower saline-water

    layers to be smaller than for upper fresh-water

    layers. The solution then should proceed withtime, solving head and salinity equations

    recursively.

    CONCLUSIONS & RECOMMENDATIONS

    The following conclusions have been drawn

    from the results of this modeling study, conducted

    to analyze the solute transport phenomenon under

  • 7/31/2019 Numerical Analysis of Groundwater-Flow and SoluteTransport under Skimming Well

    16/17

    Numerical Analysis of Groundwater-Flow and Solute-Transport under Skimming Well

    27

    skimming wells, installed in the shallow fresh

    groundwater areas:

    1. In case, the saline water mound hasalready risen, any reduction in well-

    discharge would neither decrease salinity

    of pumped water nor suppress upconing,

    but would reduce their rate of rise.

    2. In the areas having moderate values ofaquifer parameters and fresh-water

    thickness less than 30 m, the

    development of fresh groundwater

    through skimming wells is not

    sustainable. It is only the matter of time

    up to which the well will discharge good

    quality groundwater.

    3. The optimum well-penetration ratio ofthe site under study was found to be 23 %

    against the present penetration of 33 %.

    Thus, closing the lower 3 to 5 m length ofscreen and reducing well discharge by 40

    to 50 % can ensure acceptable pumped

    water-quality on sustainable basis.

    4. MODFLOW/MT3D predicted gradualrise in pumped-water salinity and

    upconing of underlying saline water, for

    all pumping scenarios. It erroneously

    provides no decline in fresh-saline water

    interface after well closure.

    5. The MT3D model does not properlysimulate skimming wells, mainly due to

    the reason that it does not take intoaccount the density difference between

    fresh and saline groundwater. For

    evaluating skimming wells, MODFLOW

    and MT3D programs need to be made to

    run, recursively, over small time-steps

    when potentials computed for various

    cells are corrected to account for

    calculated salinity levels.

    In view of the above discussion, the

    following are recommended:

    1. A thorough understanding of thehydrological properties of the aquifer at the

    specific location, where the skimming well

    facility is contemplated, must be achieved

    before the design and installation of

    skimming well;

    2. In shallow fresh groundwater areas, the well-penetration ratio should not be greater than

    20 per cent, to avoid any upconing.

    Moreover, the well should be operated at a

    reasonably low discharge and operational

    factor. The low discharges from these wells

    can be effectively utilized for irrigation,

    through pressurized irrigation systems.

    REFERENCES

    ACE & Halcrow, 2001. Exploitation and Regulation of

    Fresh Groundwater. Draft Final Report. Sector

    Policy Studies (Packages) under NDP.

    ACE,1997. Hydrological and GroundwaterMathematical Model Studies, Second SCARP

    Transition, North Rohri Pilot Project completion

    report.

    Ashraf, M., M. Aslam, M. M. Saeed, M. S. Shafique,

    2001. Effect of Intermitted Pumping on the Water

    Quality of Multi Strainer Skimming Wells.

    Proceedings 2nd National Seminar on Drainage inPakistan held in University of Agriculture,

    Faisalabad from April 18-19, 2001. pp. 200-210.Awan, N.M., 1991. Salt Water Intrusion, Centre of

    Excellence in Water Resources Engineering,Engineering University, Lahore

    Bouwer, H., 1978. Groundwater Hydrology.

    McGraw-Hill Book Company

    Chandio, B.A. and A.S. Chandio, 1992. Modeling

    Skimming Well for Irrigation and Drainage,

    Proceedings of 5th International Drainage

    Workshop, ICID-IWASRI, Lahore, Pakistan, Vol.

    2, p.

    Chiang, W.H. and W. Kinzelbach, 1996. Processing

    Modflow for Windows. A Simualtion System forModeling Groundwater Flow and Pollution. C.

    Vision Pvt Ltd 185 Ehzabeth St. Site 320 Sydney

    NSW 2000, Australia.

    Hafeez, A., Z.A. Piracha, and A. Nazir, 1986. Multi-strainer Tubewells for Skimming Top Layer of

    Fresh Water Underlain by Saline Water in the

    Aquifer, Mona Reclamation Experimental Project,

    WAPDA.

    Hunting Technical Services Ltd. and Sir M.

    MacDonald and Partners (HTS/MMP), 1965.

    Lower Indus Report, Physical Resources

    Groundwater, Vol. 6, Supplement 6.1 6.7. West

    Pakistan, WAPDA.Kemper, W. D., M. Jehangir, and D.B. McWhorter,

    1976. Skimming Well Report, Planning and

    Investigation Publication, WAPDA, Pakistan.

    Li, W.H., 1954. Interaction Between Well and Aquifer.

    Proceedings of the ASCE, Vol. 80, separate No.

    578.

  • 7/31/2019 Numerical Analysis of Groundwater-Flow and SoluteTransport under Skimming Well

    17/17

    Science, Technology & Development Vol. 30, No. 1 (January-March) 2011

    28

    McWhorter, D.B., 1975. Upconing of Salt Water Fresh

    Water Interface Beneath a Pumping Well. Journal

    of Groundwater, Vol. 13, No. 4, pp. 354-359.

    Mirbahar, M.B., A.M. Sipraw, and A.M. Rais, 1997.Performance Evaluation of Skimming Wells for

    Irrigation and Drainage. Proceedings of the

    National Congress on Impact of Drainage on

    Environment: Problems and Solutions, August 10-12, 1997, MUET, Jamshoro, Pakistan.

    Prphdpulos, S.S. and Associates. 1996. MT3D Users

    Guide.

    Saeed, M. M., M. Bruen and M.N. Asghar, 2002. A

    Review of Modeling Approaches to SimulateUpconing Under Skimming Wells-NDRDIC

    Hydrology, An International Journal, Vol. 33, No.

    2/3, pp. 165-188.

    Saeed, M.M., M. Ashraf and M. Buren, 2002.

    Diagnostic Analysis of Farmers Skimming Well

    Techniques in the Indus Basin of Pakistan.

    Irrigation and Drainage Systems. An International

    Journal, Kluwer Academic Publishers, Vol.16,No. 2, pp. 139-160.

    Sehni, B.N. 1972. Saltwater Coning Beneath

    Freshwater Wells, Water Management Technical

    Report No. 18, Colorado State University, FortCollins, USA, p. 168.

    Soliman, M.I., 1965. Boundary Flow Considerations in

    the Design of Wells. Proceedings of the ASCE

    Journal of Irrigation and Drainage Division, Vol.

    91, No. IR-1, pp. 159-177.

    Sufi, A.B., 1999. Development of Skimming Well

    Technology for Sustainable Irrigation and

    Drainage, Ph.D. Thesis submitted to CEWRE,

    UET, Lahore.

    Sufi, A.B., M. Latif, and G.V. Skogerboe, 1998.

    Simulating Skimming Well Techniques for

    Sustainable Exploitation of Groundwater.

    Irrigation and Drainage Systems, 12: 203-226.

    Zheng, C. and G.D. Bennett, 1995. AppliedContaminant Transport Modeling, Theory and

    Practice.

    Zuberi, F. A. and D.B. McWhorter, 1973. PracticalSkimming Well Design. Water Management

    Technical Report No. 27, Water Management

    Research Project, CSU, Fort Collins, Colorado.

    Zuberi, F.A. and A.B. Sufi, 1992. State of Art of

    Groundwater Exploration, Exploitation,Management and Legislation, Paper Presented in

    Expert Group Meeting of ECO on Groundwater

    Exploitation, Islamabad, Pakistan.

    ________________________________________