You are on page 1of 11
Interpretation of Pumping from Two-Zone Layered Aquifers Using a Numerical Model by K. S. Rathod and K. R. Rushton® Abstract ‘A numerical model is presented which represents both the radial and vertical flow components which usually result from pumping from boreholes in layered aquifers. This model, which is based on finite-difference approximations, runs on a ‘microprocessor system and can be used for the analysis of pumping test results. A discussion on the interaction between the radial and vertical flow components is followed by a description of the model. A computer program in BASIC and a typical computer output are provided. The application of the model for the analysis of a pumping test in a weathered-fractured aquifer is described. Three case studies are reviewed where the numerical models used to gain a greater understanding of the aquifer flow processes. Introduction ‘The importance of the combined effects of radial and vertical flows in aquifers due to pumping from boreholes is ‘becoming more widely recognized. Failure to consider verti- cal flow components often occurs because many of the available methods of analysis assume predominantly radial flows. However, for leaky aquifers, layered aquifers or unconfined aquifers, vertical flow components can be of sgreat importance. Vertical flow components also arise from partial penetration of wells or from alternating slotted and solid casing of the borehole assembly Many of the standard analytical methods of pumping test analysis assume that the flow is radial. Other methods ‘which allow for vertical flows, such asthe leaky aquifer type curves (Jacob, 1946; Hantush and Jacob, 1955) or the unconfined aquifer theory of Neuman (1972), provide important techniques for including the vertical flow compo- nents but they are unable to represent the complexity of ‘many of the multilayered situations which occurin the field. ‘Numerical methods based on discrete-space/discrete- time approximations can provide a greater flexibility Recently, Walton (1987) presented a program for a numeri- cal model of radial flow to a pumped well that was based on earlier work by Rushton and Redshaw (1979) and Rathod and Rushton (1984), Another numerical method which has ‘been widely used in practical applications is based on a “two-zone” approach (Rushton and Booth, 1976). In the two-zone approach, the conceptual aquifer system is approxi- ‘mated as two horizontal permeable zones with less perme- able intermediate layers. When this approximation is adopted, it is possible to represent both the radial and vertical components of flow. More detailed models are ‘School of Civil Engineering, University of Birmingham, Birmingham BIS 2TT, United Kingdom. Received January 1990, revised June and October 1990, accepted Ociober 1990. Discussion open until January 1, 1992 Vol. 29, No, 4~GROUND WATER— July-August 1991 available in which the aquifer system is divided into asignifi- cant number of radial and vertical mesh intervals (Taylor and Luthin, 1969; Neuman and Witherspoon, 1971), but these approaches require considerable computational effort. There may not be sufficient field data to justify the use of these detailed models for pumping test analyses. ‘The two-zone model has proved to be useful in under- standing the flow mechanisms in a wide variety of aquifers. ‘The purpose of this paper's to draw together the experience ‘gained from these studies and to present a description of the ‘model which includes a full range of the important features. ‘The model includes the effects of radial and vertical flow in an aquifer system having two permeable zones and inter- mediate and overlying low permeability layers. Other features include confined storage, specific yield and delayed yield, well losses, and different conditions at the outer boundary. A computer program using a discrete-space/ discrete-time approach is developed and a version in BASIC. is available which can be run on a microprocessor. A. FORTRAN version of the program is also available. The preparation of datais described, and the use ofthe model for analyzing a pumping test in a weathered-fractured aquifers discussed. The application of the model in three other situa- tions is outlined Radial and Vertical Flow Components in Zoned Aquifers ‘A number of different geohydrologic conditions that ccan affect the dynamic flow of water when pumped from a borehole are illustrated by sketches in Figure 1. Figure I(3) refers to aconfined aquifer in which the borehole has asolid casing through the upper part of the aquifer but is open in the lower part of the aquifer. Significant vertical compo- nents of flow will occur in the vicinity of the borehole. The example of Figure 1(b) refers to an unconfined aquifer with solid casing in the borehole a substantial dis- tance below the water table. Vertical flows occur as the water released from storage at the water table moves downwards towards the open portion ofthe borehole. Failure to repre- 499 pumped pumped water table low nydeavi poe agulter conductinty aquifer with borehole only open in lower part of the aquifer, (b) unconfined aquifer with solld wo! water table in Dhaai acter Overiying layer Upper zone Me layer a Fig. 1. Typical examples of flow towards a pumped borehole when vertical flow components are significant: (a) confined ing inthe upper part of the aquifer, c) two aquifer zones separated by alow permeability ayer with upper zone unconfined, (d) layered aquifer with well ‘pen to both upper and lower aquifer zones. sent the vertical lows will mean that an important flow mechanism is ignored Figure 1(0 illustrates an unconfined aquifer in which there is layer of material with low hydraulic conductivity between the upper aquifer and the lower aquifer. The bore- hole draws water from both the upper and lower zones but has solid casing through the low conductivity layer. The flow patterns will be complex with both radial and vertical com- ponents in the two aquifers but with predominantly vertical flows through the low permeability layer. One important feature is the distribution of the flow into the borehole between the upper and lower zones; this can only be deter- mined from an analysis which includes both radial and vertical components of flow. ‘The final example, Figure 1(d), refers to a system in which the upper aquifer zone is overlain by a less permeable layer and a phreatic aquifer; vertical flow can occur through this less permeable layer. In addition, there is an inter- mediate (middle) low permeability layer separating the upper and lower aquifers. Water which enters the upper aquifer from the overlying layer may flow towards the open section of the borehole in that zone, or it may be drawn through the intermediate layer into the lower aquifer zone and then to the borehole. This example reinforces the importance of including both radial and vertical compo- nents of flaw. Description of Two-Zone Model Parameters of the Model Although four examples have been selected including confined, unconfined, and multilayered aquifers, they can be represented adequately by the two-zone model illustrated in Figure 2. In its general form the model consists of four horizontal zones (or layers or strata), each designated by two letters: 500 © the Overlving low permeability (leaky) layer, .... (OL) © the Upper aguiter zone, 7 : (UP) © the Middle low permeability layer (MD) © the Lower aquifer zone. (Lo) Al distances to the interfaces between the zones and layers are measured vertically downwards from the rest- water level, RWL, The top (TP) and base (BS) of the upper and lower aquifer zones are defined by TPUP and BSUP, and TPLO and BSLO with the top of the overlying layer written TPOL. ‘The two-zone model uses discrete-space diserete-time approximations. The radial dimension is divided into a ‘mesh which increases logarithmically six mesh subdivisions for a tenfold increase in radius are usually sufficient) from the borchole radius to an outer boundary. The mesh for each of the permeable zones is terminated at the outer boundary cither by a zero-flux or a zero-drawdown condi- tion. If there isan effectively infinite source of water such as deep lake whichis in contact with the upper or lower zone, a zero-drawdown condition may be appropriate. The alter- native zero-flux condition is frequently used; it corresponds. roe SETS | Perea eee Or, LE ZoLe }Ovening layer Fig. 2. Conceptual features of the two-zone model to a physical boundary to the aquifer or to a reduction in transmissivity; alternatively, the zero-flux boundary can correspond to the midpoint between neighboring pumped boreholes (Holt and Rushton, 1984) or a ground-water divide. For certain problems, the location and type of outer boundary condition may be different in the upper and lower zones. For example, there may be zero drawdown at a specified radial distance in the upper zone with the lower zone extending to a greater radius, This can be included by setting an artificially high storage coefficient of 100,000 at the upper zone node which corresponds to this recharge boundary, thereby ensuring that the drawdown is negligibly small Conditions must also be defined at the face of the borehole. The borehole is likely to have solid casing through the low permeability layers. Itis aso possible that either the upper or lower aquifer zone may have solid casing, alterna- tively, water may be drawn into the borehole from both zones, When water is drawn from a single zone, this can be included by allocating the total discharge exclusively from ‘that zone. When the water is drawn from both of the zones, the drawdown at the well face is the same in each zone with free vertical transfer within the borehole, Well storage effects are automatically included by extending the mesh into the borehole and setting the storage coefficient within the bore- hhole equal to unity (Rushton and Redshaw, 1979), Close to ‘the well and at the face of the well, there is often a rapid decline in the ground-water head. This rapid decline occurs due to well losses (Clarke, 1977; Rushton and Rathod, 1988), which can occur due to changed hydraulic conduc- tivity following drilling, the effect ofthe gravel packs or the ‘well screen, turbulent losses, or the effect of a seepage face (Sakthivadivel and Rushton, 1989). Well losses are repre~ sented as a change in the hydraulic conductivity close to the well A logarithmic increase in time increment is also used; ten time steps for a tenfold increase in elapsed time are adequate, although for large-diameter wells, it is advisable to double the number of time steps per given time period. ‘The initial time step is 10” day. This value is chosen so that the initial well drawdown is less than 1 mm; the time step must reduce to this low value for each change in pumping rate, Pa Parameter values must be assigned to each of the zones; these parameters may vary with the radial distance from the {est borehole, Up tofive parameters ae required to describe conditions in each of the four strata; these are layer thick- ness, radial and vertical hydraulic conductivities, and con- fined and unconfined storage coefficients. The parameters used in the computer program are listed in Table 1. For example, PERMVLO refers to the hydraulic conductivity, ie, (PERM)eability in the (V)ertical direction for the (LO)wer zone. For the overlying and middle layers itis assumed that the vertical hydraulic conductivities are at least two orders of magnitude less than the radial hydraulic conductivities ofthe upper and lower aquifer zones. Conse- quently, the low through the overlying and middle layers is predominantly vertical and can be described adequately by specifying the thickness and vertical hyéraulic conductivity of these layers (Neuman and Witherspoon, 1969). The radial hydraulic conductivity and storage coefficients for these layers are not required; this is indicated by N/R in Table [ In this formulation the storage coefficients of the low permeability layers are not represented directly. A detailed study of the approximations inherent in the methods of analyzing leaky aquifers is presented by Neuman and Witherspoon (1969); in certain situations the effect of neglecting the storage effects in the low permeability layers can be serious, but in the examples discussed in this paper the effects are not very large. For the lower aquifer zone the important parameters are the thickness, the radial and vertical hydraulic conduc- tivities, and the confined storage coefficient. For the upper aquifer zone, two possibilities should be considered; if the zone remains confined the confined storage coefficient for this zone is used. However, when the drawdown in the upper zone reaches the base of the overlying layer, water-table conditions apply and the unconfined storage coefficient and, if delayed drainage occurs, a delayed yield index must be defined, Delayed drainage may occur as the water table falls. Boulton (1963) represented the delayed release of water at the water table in terms of delay index, a. Boulton’s delayed yield is incorporated in the numerical model in terms of an effective storage coefficient, which equals the confined storage coefficient at the start of pumping and increases to the specifi yield at late times, plus an effective for Numerical Model Hydraulic conductivity Storage coefficient ayer Thickness Raed «Vertical Trconfned Overlying TPUP-TPOL NR ~PERMVOL Nik Upper BSUP-TPUP. PERMRUP, PERMVUP, * or BSUP-DUP()* PERMRUP, PERMVUP, SUNCNUP Middle ‘TPLO-BSUP NR PERMVMD NUR Lower BSLO-TPLO PERMRLO, PERMVLO, N/R This parameter is not required for this zone or layer. SCONLO NIR = These conditions apply when the upper aquifer zone becomes unconfined; DUP(N) is average drawdown between nodes N and N + 1in the upper aquifer zone. 4 A value for this parameter must be provided but it will only be used ifthe aquifer, which was initially confined, becomes partially ‘unconfined. S01 Table 2. Parameters for the Example of Figure 1(b) Zone “Hydraulic conductivity Storage coefficient layer Thickness Radial Fercal ‘onfned Tneonfned Overiying 00 NR 00 oo Upper BSUP-DUPO) PERMRUP PERMVUP ScoNUP SUNCNUP. Middle oo NR PERMVMD. NR NR Lower BSLO-TPLO, PERMRLO, PERMVLO SCONLO N/R This parameter isnot required for this Zone or layer. recharge which represents the delayed drainage above the water table. For further details see Rushton and Redshaw (1979). During the computation it is possible in the upper aquifer zone for conditions to change between confined and ‘unconfined at different nodal points; this requires that both confined and unconfined storage coefficients are read for this zone, and the program automatically selects the appropriate value depending on whether the node is con- fined or unconfined. Recharge can be included in the model whenever appropriate; it is similar to the water released from storage and has units L/T. Recharge values are stored in the array RECHO. With so many parameters, itis not practicable to design an optimization method to identify aquifer parameters, hence a trial and error method is used. Representation of Different Flow Systems The numerieal model is eapable of representing all the flow systems of Figure I. For the example of Figure (a) in which the confined aquifer contains a borehole cased through the upper part of the aquifer, the vertical hydraulic conductivity ofthe overlying layer is set to zero to represent, the nonleaky confined conditions, an the middle low per- :meabilty ayer is positioned atthe base ofthe solid easing and given zero thickness. In addition, the specific yield ofthe upper zone is set to zer0, and all the discharge is specified from the lower zone. With the unconfined aquifer of Figure 1(b), the condi- tions are similar to the frst example but the overlying layer has zero thickness. This is specified by setting the top of the overlying layer, TPOL, at the same elevation asthe top of the upper aquifer zone, TPUP, with zero vertical hydraulic conductivity, PERMVOL. The saturated depth of the upper aquifer zone decreases with continued pumping; therefore the unconfined storage coefficient must be speci- fied (plus a delay index, ALPHA, if appropriate). Table 2 indicates how parameters are selected for this problem for Which there is no overlying layer. For the unconfined aquifer of Figure 1(e), the middle low permeability stratum is included as a layer of specified thickness and vertical hydraulic conductivity. The upper and lower zones have different hydraulic conductivities, and the total discharge is withdrawn from the two zones. By connecting the nades which represent the upper and lower zones within the borehole to ensure that they have the same drawdown, the share ofthe discharge between the two zones is determined automatically as the calculation proceeds. ‘Well loss parameters can have an important effect on this 502 NIR distribution of discharge between the two zones; the case studies discussed later in this paper illustrate how the well loss parameters are estimated. All of the parameters will have nonzero values for the fourth example, Figure 1(@). This example will form the basis of a test problem which is described in a later section, Derivation of Discrete-Space/ Discrete-Time Equations ‘There are two methods of deriving the coefficients of the numerical model: in one the fnite-ifference form of the appropriate differential equation is used while the second, which is discussed below, is based on a lumping approach, Figure 3. Each method leads to the same equations. The lumping approach is used below; for further details of the finite-ifference approach, see Rushton and Redshaw (1979) Consider node n in the lower zone; water can move ang Ta es owen ee ode DOW ans i Fig. 3. Derivation of numerical mod. flow balance in lower zone, (b) equi resistances, (a) components of lent hydraulic. towards this node in three directions: radial flow through the ower zone from node n+ 1; radial flow through the lower zone from node n ~ I; and vertical flow through the middle layer from node n in the upper zone. In addition, water is released from storage due to ‘changes in the drawdown with time. The calculation of each of these components is described below; note that the vari- able used in these calculations is the drawdown below the initial rest water level Using the Theim equation for radial flow in an aquifer, the radial flow from node n + 1 to node n, Figure 3a), equals QL Ope = —2zHMEDLO ss — DLO,} 1 ln /ta) o where kr = radial hydraulic conductivity; m = saturated thickness; DLO,, DLOg+i = drawdowns at nodes n+ Lin lower zone; and ts, rast = radial distances to nodes n and n+ 1. The negative sign is required sine a larger drawdown at node n+ [indicate a flow away from node To ensure that there isa small mesh spacing close to the borehole, with the spacing increasing away from the bore- hole, logarithmicincrement is used. The logarithmic spac- ing requires the use of an alternative radial variable a, such that, a= In(2), This leads tothe relationship re= Ba = In(tas te) Substituting fortn (ot ta) equation (1) can be written ato, 2m(DLOn+1 — DLOn) @ me Aaj (Kem) iilarly, radial flow from node n ~ 1 is given as ~2r[DLOy-1 = DLOw) QLO,-1 = arene Q) aj em) ‘The vertical flow through the middle layer from noden in the upper zone is Aks{DI 4 where ky = vertical hydraulic conductivity between the two Zones; DZ = vertical distance through which flow takes place; and A= annular plan area represented by the node= QataAr = 2m’ Aa. ‘Substituting for A, equation (4) can be writen as —2nAa{DUPs ~ DLO,} 5 DZi(krs) : “The fourth component, ater released from confined storage duet the change in drawdown at this node, equals AS.[OLDDLO, — DLOn} , = TASHOLDPLO, = PhOn} 6 Q at (6) where Se = confined storage coefficient at the node; At = time step over which the change in drawdown occurred; OLDDLO = drawdown at this node at the end of the last ime te; and A= annular area associated withthe node as described above Replacing for A, equation (6) becomes 2r4a{OLDDLOy ~ DLOs] Aty/(Sera’) Considering the flow balance at this node, all the four components must sum to zero. From equations (2), (3), (3), and (7), Q= o [DLOo+1 = PLOn} , 2#[DLOn-1 ~ DLOn] ‘Ba/(kem) ‘Aa/(kem) 2nAa[DUP, ~ DLOg] DZ (kets) 2rAa[OLDDI _ oe @) tere) Dividing equation (8) by 27Aa throughout, and rearranging, [DLOn+1 ~ DLOs]_, [DLOs-1 ~ DLOn] Aeydem am) {DUP,~ DLO_] _ =[OLDDLO,~ DLO} ig, DZi(kite) ayer) The denominators ofthe expressions in equation (9) will be defined as: radial hydraulic resistance: HLO() = Aa’ /(kem); V() = DZ / (Kot), TSLO() = At/(Sern’) ‘Therefore, each of the flow components is calculated using the differences in drawdowns and an equivalent hydraulic resistance, Figure 3(b). An identical equation is obtained by developing a finte-difference approximation to the govern- ing Now equations (Rushton and Redshaw, 1979). A similar equation for the corresponding node in the upper zone can be obtained; it contains an additional term to represent vertical flow through the overlying layer to the upper zone of similar form to the term for vertical flow through the middle layer, equation (4). Also, the storage coefficient in this case may have components due to both confined and unconfined storage and could include the effect of delayed yield (Rushton and Redshaw, 1979). vertical hydraulic resistance: time [storage coefficient Solution of Equations In terms of the variables used in the computer program, equation (9) can be written as, [DLOW+)~DLOW)] , [PLOG—)—DLOWN)] HLOW) HLOW-1) LOt {OLDDLO(N)~ DLO(N)} TSLO() (a) In the preceding equation (9a), all the drawdowns except OLDDLO(N) are at the current time step and are unknown, Defining: 503 ALO(N) = 1/HLO(N— 1), CLO(N) = 1/ HLOW), EM)=1/V), BLO(N) =~[ALO(N) + CLO(N) + E(N) + 1/ TSLOWN)}, and FLO(N) = —OLDDLO(N)/TSLO(N), and rearranging, ALO(N)- DLO(N ~ 1) + BLO(N) - DLO(N) + CLO(N) + DLO(N + 1) + E(N)* DUP(N) = FLO(N) (10) A corresponding equation for node N inthe upper zone is AUP(N) - DUP(N ~ 1) + BUP(N) - DUP(N) + CUP(N): DUP(N + 1) + E(N}- DLO(N) = FUP(N) -) Note that in (ID, the variable BUP () includes an addi- tional coefficient for leakage through the overlying layer to ‘the upper zone, and also the Time/ Storage resistance has to be modified to include any effects of delayed yield and uunconfined/confined storage. The coefficient FUP() includes any contribution from recharge or flow through the ‘overlying layer to the upper zone. Equations (10) and (11), taken simultaneously, describe the flow balance at node N in the lower and the upper zone. Solutions to these equations with the appropriate boundary conditions are obtained using a Gaussian elimination rou- tine, Due to the wide rangein magnitude ofthe coefficients, itis essential to use double precision variables in the com- puter program. The numerical model can be used for a ‘number of specific problems for which analytical solutions are available. The agreement between the analytical and ‘numerical solutions is good with differences in drawdown usually less than 1% (Rushton and Redshaw, 1979: Grout, 1988). A flow balance is automatically achieved in the numerical solution, However, inaccuracies may occur when complex field problems cannot be adequately represented by the conceptual two-zone model Figure 4 contains a flow chart of the program; the program will run successfully using interpretive BASIC software, but the computational time is farlessifa compiled BASIC is adopted. Copies of the program which include both BASIC and FORTRAN versions are available from the authors. Applications Description of Representative Example A representative example, Figure 5, i used toillutrate the application ofthe numerical model. The example relates to situation similar to Figure I(@) with a two-zone aquifer ‘overlain by aleaky layer and with a low permeability middle layer separating the two productive zones, Reference should also be made to Table 3 which lists the aquifer parameters in thesame form as Table 1, and to Figure 6 which contains the input data for this example. ‘The production borehole has & radius of 0.2 m and is ‘open to each of the productive zones. A zero-flux boundary 504 Fig. 4. Flow chart of computer program for the two-zone model. condition is enforced at 5,000 m limiting the area from which the borehole can withdraw water. The test pumping starts from a rest condition with a pumping rate of 1,500 ‘md for 0.25 day; recovery continues fora further day to an lapsed time of 1.25 days. ‘Aquifer parameters are recorded in Table 3; the ft ceeded Fig. 5. Details of typical example ble 3. Parameters for the Example of Figure 5, Dimensions m or md Zone Hidraulicconducivity ‘Storage coins layer Thickness adkal Vertical Confined Tnconfned Overiving 300250 NR 001 NIR NR Upper 650.500 30 12 0.0008 2 or 6S0-DUPU* 30 12 0.0008 os Midate 0045.0 Nik us NR NIR Lower 1250900 10 as 0.0006 Nik NjR This parameter isnot required for this zone or layer. * These conditions apply when the upper aquifer zone becomes unconfined; DUP(N) is average drawdown between nodes N and N+ 1 in the upper aquifer zone. abstraction rate is not high enough to cause the ground ‘water head in the upper zone to fall below the base of the ‘overlying layer; consequently, delayed yield will nt operate Well loss factors for the upper and lower zones are set at 9.0 and 6.0, respectively; the radial hydraulic resistances adj cent to the borehole are multiplied by these factors to represent the reduced hydraulic conductivities. Four obser- vation borehole locations are identified at 10, 30, 100, and 300 m from the pumped borehole ‘The input data together with explanatory notes are contained in Figure 6, LEVPRN is a parameter which is used to control the amount of printout produced by the program, When LEVPRN = |, little information about aquifer parameters is printed. Drawdowns at observation ‘boreholes are printed at the end of each time step and a ful summary including the distribution of flows is produced at the end of each phase. When LEVPRN = 2, in addition to the printout of LEVPRN = 1, the aquifer parameters for ceach node in tne aquifer are printed. When LEV PRN = 3 addition to the printout with LEVPRN=2, afullsummary at specified times TP() is produced. igure 7 contains part ofthe output from the program ‘at Level I, Information is provided for times of 0.00035 day (0.5 min) and greater. The first line of output for each time step refers to the flow into the borehole from the upper zone ‘QUP, from the lower zone QLO, the contribution from well storage and the quantity of water moving vertically through the overlying layer into the upper aquifer zone; all the ‘quantities are in m'/d. The second and third lines contain information about the drawdowns (in meters) in the upper and lower zones at the borehole, at the four observation borehole locations and at the outer boundary. ata Descererton 20, 5000.0 RUELL, RK: welt radtue and aaeioum radfus 10,38.0,50,0,65.0,90.0,125.0 RWL, TPOL, TFUP, BSUP, TPLO, BSLO: see noce 2 0, 30.0, 100.0, 3000 DOBS():" radial distances 1, 2.0, 0.0, 0.0 TPQ: cles for decatlod 0125) STOP: punping ra 135 rotor: QeuNF-O 20 rst0e: ve QeumP for JFIX'~ 1 recharge boundary, for JFIL = 2 no-flow boundary feo note 1 his can change parameter values ac node 8 T's 0 terminates’ input of non-standard values O eerninates Input of non-standard values of PERKMD T= 0 terminates inpue of non-standard values 1.2, 0.0008, 0.08 PERGUP, PERAVUP, SCOSUE, suNCHU: 112} 0-0008, 0:03 x, PERMRUP, PEaMNUP! scoNUE, SUNCKDP: 0:7! 0:0001, 9:02 T PemiRUP; PEMAVUP, ScONUe, sUNCNUP: PEUIMD! “see nose 2.5, 0,0006 PERIRLO, PERALD, SCONLD: see noce 7 1.1; 0:0001 1, PERRO, PHOVD, ScoNLo, 0, 6.0 nips, HLOSSUP, WLDSSLO: see noce 2 TELL? see note 7 NOES: number of observation borehotes £ observation boreholes eine out when LEVPRIAI; zeros are ignored ‘nest time TSTOP ster zecovery phace ‘signifies ond of program zn ‘those variables have been defined in the text of the paper: nitisLly, the values axe read inco single ‘variables and then transferred to the array variables PERGUP() ee Note 2 ALIA t4 che delay index, ULOSSUP ane ULOSSLO are well loss factors which Note 3 IWELL defines sbstrsetion conditions: WELL = 3 Note 4 LEVPRN defines the detail of the print out; 1 Fig. 6. Conceptual representation and input parameters for 1£ tr 10 negative chore 19 no delayed yield. fc je resistance £9 flow at well face. eraction only from lover zone: fraction only from spper Zone: Fully penetrating well, abseraceion from both zones sninimun to 3 = waxtoun, “example problem. sos Fig. 7. Portion of program output for “example problem.” 4 w@ | ] vou rom as tower zone tt Application to Weathered-Fractured Aquifer ‘A study of a weathered fractured granitic aquifer will bbe used to illustrate the use of the two-zone numerical ‘model; the approximate thicknesses of the weathered and fractured zones are indicated in Figure 8(a). A borehole drilled into the fractured zone ofa granitic aquifer with solid casing in the weathered zone was test-pumped using air- lifting. Drawdowns are recorded in Figure 9for the pumped borehole and for two shallow observation boreholes in the ‘weathered zone at 20 m from the pumped borehole [Figure 8(a)]- Drawdowns were obtained for the 7.5 hours (0.31 day) during which the pumping rate was 288 m’d and for 3.5, hours (0.15 day) ofthe recovery; further information about the test can be found in Rushton and Weller (1988). ‘An important feature of the field results is that the levels in the observation boreholes in the weathered zone continued to fall during the initial part of the recovery although the final field readings suggest a leveling-off this indicates the possibilty of vertical flow ftom the weathered into the fractured zone. Consequently, in the numerical ‘model the upper zone corresponds to the weathered zone and the lower zone corresponds to the fractured zone. ‘The numerical model requires parameter values to be defined and provides predicted drawdowns due tothe speci- fied pumping regime, Whenever possible, existing type- curve methods are used to obtain first estimates of the aquifer parameters; single-well responses or slug tests can also be helpful (Cooper et al, 1967, Hvorsev, 1951). Alter- natively, typical values, such as those presented in a helpful ) ay wary sand ge eA mainly sand | Siay wih sand i Fig. 8. Examples of the practical use of the two-zone model: (a) weathered-tractured aquifer, (b) layered aquifer in Yemen, (c) unconfined sandstone aquifer, (4) alluvial aquifer. 506 Fig. 9. Results from the two-zone model of the weathered- fractured aquifer: comparison between field and modeled 0.0 THEN QF=232.0 ~ 4.58+ (DLO(2))*3.0 ELSE QF=0.0 395 QABST=(QF+QPUMP) /(2.0*PI*DELA) ‘The results from this continuously varying rate from the pumped well are shown by the broken line in Figure 9; the results for the observation boreholes remain almost unaffected. Instead of using the cube of the drawdown in the expression for the reduetion in abstraction rate, adequate results can also be obtained using the square of the drawdown, Having obtained a reasonable match for the short-term pumping est, the numerical model was used to estimate the response over a typical growing season (120 days). Long- term predictions based on a model which has been verified only for a short period should be treated with caution. The results can only provide an indication of likely trends, and it is important to check these trends against observed field responses. Each day the pumping schedule is 288 m'/d for 10, hhours and recovery for 14 hours. Itis assumed that there are a number of boreholes in the aquifer so that the area from which this borehole can draw water is 500 m 500 m; this is represented as an impermeable boundary at 282 m. Figure 10 shows the predicted drawdown in the pumped borehole, at the observation boreholes in the weathered zone at 20 m, and at the outer boundary. Significant dravdowns occur both in the fractured and weathered zones. After 120 days the pumped dravidown is 8.3 m with a drawdown at the ‘outer boundary of 4.0 m. Unless there issignficant recharge, the ground-water heads will fall year by year. In the area ‘Table 4. Parameters for the Weathered-Fractured Aquifer, Dimensions m or mid Zone layer Thickness Rahal Overlying 0.0- 00 — Upper no DUPO* 50 Midale t0130 NIR Lower w0130 40 HURT pratt vue fovthi ane npr Hydraulic conductivity ‘Storage coefficient Confined Unconfined * These conditions apply when the upper aquifer zone becomes unconfined; DUP(N) is average drawdown between nodes N and N+ 1 in the upper aquifer zone, 307 Fig, 10. Predicted drawdowns for pumping cycle repeated for 120 days: (a) results for the first 10 days, (b) results for full 120 days. where this test was carried out, there has been a steady decline in ground-water heads with the result that the weathered and fractured zones can no longer provide a reliable supply Further Examples ‘Three further examples of the application of the two- zone numerical model are discussed below. Additional case studies including gravel, chalk, and limestone aquifers are described by Rushton and Booth (1976), Connorton and Reed (1978), and Rushton and Rao (1988) Layered Aquifer in Yemen The frst example relates to a layered aquifer in Yemen (Sutton, 1985; Davey, 1989) in which unconsolidated Wadi deposits overlie Cretaceous sandstones. Between these two strata is a well-cemented conglomerate, Figure &(b). Pro- duction boreholes have been drilled to tap each of these aquifers ‘The numerical model which is described in this paper ‘ean represent the two main aquifer zones. The important vertical flow between the zones is dominated by the influ- ence of the conglomerate, Since there is no overlying layer, the Wadi deposits are unconfined and a delay index is included. A seepage face forms atthe well face of the Wadi deposits. Using a calibration procedure to match the field and modeled drawdowns in the pumped well during the 508 pumping and recovery phases, well loss factors of 12 were required in each zone for adequate agreement in the pumped borehole, The numerical model was used to interpret a wide range of pumping tests in this aquifer (Sutton, 1985); an important finding was that the low vertical permeability of the conglomerates resulted in water being drawn from {greater radial distances through the sandstone aquifer of the lower zone. Uncontined Sandstone Aquifer The Lower Mersey Sandstone aquifer contains exten- sive horizontal marl bands and a number of significant fissures (Bliss and Rushton, 1984). Pumping tests were car- ried out with shallow and deep piezometers, and it was found that different hydrographs were obtained with depth (Rushton and Howard, 1982). There is no distinctive divi sion within the aguifer unit, and therefore an arbitrary horizontal division was defined at half ofthe effective satu- rated depth, Figure 8(¢). From a series of calibration runs, the radial hydraulic conductivities of the upper and lower zones were deduced tobe 1.2 md. The vertical permeability is strongly influenced by the mari bands resulting in low average vertical hydraulic conductivity of 0.20 md. Draw

You might also like