Multi-Level Slug Tests in Highly Permeable Formations: 1. Modification of The Springer-Gelhar (SG) Model

You might also like

Download as pdf or txt
Download as pdf or txt
You are on page 1of 12

Journal

of
Hydrology
ELSEVIER Journal of Hydrology 204 (1998) 271-282

Multi-level slug tests in highly permeable formations:


1. Modification of the Springer-Gelhar (SG) model
Vitaly A. Zlotnik, Virginia L. McGuire I
Department of Geosciences, University o f Nebraska-Lincoln, 214 Bessey Hall, Lincoln, Nebraska 68588, USA
Received 3 February 1997; revised 27 October 1997; accepted 29 October 1997

Abstract

A multi-level slug test model and a method for the evaluation of vertical profiles of hydraulic conductivity in highly
permeable formations are developed. A double-packer system is employed to estimate local hydraulic conductivity. Depend-
ing on the formation and double-packer system parameters, the water level recovery in the tested well can exhibit a monotonic
or oscillatory response. To discern information on aquifer properties from artifacts introduced by the measurement system, the
theory of flow in an aquifer and a double-packer system is developed. The mathematical model incorporates features of the
Springer and Gelhar (1991) model and reduces to the Bouwer and Rice (1976) model in a special case. The model involves
equations of momentum and mass conservation for the double-packer system with quasi-steady well-aquifer interaction
equations. The method is uniformly applicable for both monotonic and oscillatory well responses and can produce profiles
of hydraulic conductivity for the tested well. The criterion is given to determine the type of well response for given slug test
conditions. © 1998 Elsevier Science B.V.

Keywords." Hydrogeology; Slug test; Hydraulic conductivity; Modeling; Alluvial aquifer

1. Introduction spacial vertical and radial dimensions on the order of


a perforated screen length.
The slug test is an important tool for characterizing Commonly, water level recovery in a tested well
the local hydraulic conductivity K in heterogeneous exhibits both monotonic (non-oscillatory) and oscilla-
aquifers. These methods are based on invoking instan- tory responses.
taneous initial water level displacement in the tested
well and observing water level recovery after the 1.1. Non-oscillatory response
release o f the initial aquifer perturbation. The hori-
zontal hydraulic conductivity Kr o f the aquifer can Analysis methods for slug tests with non-
be derived from an interpretation o f the water level oscillatory slug responses employ two different
recovery (see the review by Butler et al. (1996)). This approaches. One approach by Cooper et al. (1967)
value is the average Kr for the aquifer volume with includes both compressibility and Kr. However, it is
limited to a simple well geometry and hydrostrati-
Current address: U.S. Geological Survey, 100 Centennial Mall graphic conditions because it considers only a fully
North, 406 Federal Bldg., Lincoln, NE 68508, USA. penetrating well in a uniform confined aquifer. The

0022-1694/98/$19.00 © 1998 Elsevier Science B.V. All rights reserved


Pll S0022- 1694(97)00 128-5
272 V.A. ZIotnik, V.L. McGuire/Journal of Hydrology 204 (1998) 271-282

method produces values of Kr and specific storage Rice (1976) approaches are frequently recommended
averaged over the aquifer saturated thickness for for highly permeable formations.
both low and high conductivity formations, confined
and unconfined conditions, e.g. (Neuzil, 1982; Belitz 1.2. Oscillatory response
et al., 1995). The specific storage is usually deter-
mined with an accuracy of an order of magnitude. Oscillatory water level recovery occurs due to the
The anisotropy of the aquifer material cannot be inertial effects of a moving water column inside a well
determined by this test. The model is relatively insen- screened in highly permeable formations (Cooper
sitive to changes in storativity compared to changes et al., 1965; van der Kamp, 1976; Krauss, 1977;
in K. Kipp, 1985; Shapiro, 1989; Stone and Clarke, 1993).
Another approach by Hvorslev (1951) and Bouwer Oscillatory water level behavior is also observed as a
and Rice (1976) emphasizes the role of well geome- response to periodical earth tide and atmospheric
try, screen size, and hydrostratigraphic conditions loading (Bredehoeft, 1967; Rojstaczer, 1988; Ritzi
near the well screen. This approach produces K-values et al., 1991). All these studies were carried out for
averaged over a smaller scale provided the anisotropy fully penetrating wells. Analysis of oscillatory beha-
of the aquifer K is known (Zlotnik, 1994). This vior in fully penetrating wells is relatively complex.
method is designed for evaluation of local K using Typically, only approximate or numerical solutions
short-screen wells or partially penetrating boreholes, are available. Extension of existing models to short-
and it is commonly used in conductive formations. screen wells requires even more computational effort.
These approaches are not irreconcilable. It was Currently, the need exists for techniques to deter-
shown conceptually (Dagan, 1978) and numerically mine small-scale K at thousands of locations. In some
(Widdowson et al., 1990) that one can safely neglect cases, the slug tests in short-screen wells are the only
the compressible properties of the high permeability technique available (Hess et al., 1992).
aquifer when storage coefficient values and the To our knowledge, only two studies have been
volume of displaced water are small. This concept is published on oscillatory responses for short-screen
very effective in sand and gravel aquifers. Dax (1987) wells. The first study by Springer and Gelhar (1991)
demonstrated that models by Cooper et al. (1967) and derived a slug test model for wells of uniform dia-
Bouwer and Rice (1976) produce similar responses meter by combining a simplified equation of pipe
if well recovery curves are plotted for a certain hydraulics (Kipp, 1985) and the Bouwer and Rice
range of time and for given values of storage coeffi- (1976) model for water exchange between the well
cient. (In aquifers with low to moderate conductivity, and the aquifer. The model was used for slug test
this approach must be used with caution (Hinsby analysis in more than 338 wells with different screen
et al., 1992).) lengths at the Cape Cod site, where about one third
Widdowson et al. (1990), Hyder et al. (1994), of well responses were oscillatory.
Hyder and Butler (1995), and Brown et al. (1995) Another study on inertial effects was published
presented numerical models for simulating slug tests independently by McElwee et al. (1993); McElwee
in a broad range of conditions considering short et al. (1994). Their theoretical derivations were
well screens, aquifer compressibility, and skin. based on analysis of turbulent flow in the well and
These comparative analyses of the methods of Cooper the Hvorslev (1951) model and were supported by
et al. (1967), Hvorslev (1951), and Bouwer and Rice the testing of several wells. The authors also encoun-
(1976) indicate that the approaches produce remark- tered fully oscillatory regimes or significant devia-
ably similar hydraulic conductivity estimates in tions from typical responses for the Hvorslev (1951)
unconfined aquifers with small values of storage or Bouwer and Rice (1976) models.
coefficient and high K for the tests at distances of at
least a few screen lengths from the water table. Con- 1.3. Multi-level slug tests
sidering the uncertainty associated with identification
of the averaging scale for the slug test and the high Only one K value is associated with the slug test in
variability of K, the Hvorslev (1951) and Bouwer and a short-screen well. By using a double-packer system,
IdA. Zlotnik, ILL. McGuire/Journal of HydrologT" 204 (1998) 271-282 273

a K-profile can be obtained for multi-level slug tests in well; and multi-level slug test results from 11 wells
a long-screen well and this drastically increases the in 156 locations at the MSEA site (Shelton,
amount of information collected from each well. Nebraska).
Melville et al. (1991) demonstrated that a double-
packer system could significantly increase the spacial
resolution K~ for slug tests in sand and gravel aquifers. 2. Theory
Kr-profiles were obtained from several wells 0.15 m
in diameter with screens up to 6.0 m long. The double- Consider a partially or fully penetrating well that is
packer system divided each tested screen into ten screened between two given elevations (Fig. 1). An
0.6 m intervals which increased the number of test arbitrary section of the screen can be chosen for test-
locations by an order of magnitude without installing ing of aquifer hydraulic properties using a double-
new wells. All the slug test responses were non- packer system. The radius of the borehole or well
oscillatory and were interpreted using the Bouwer screen r~, the radius of the riser pipe rr, and the radius
and Rice (1976) method. Simulations of multi-level of the access pipe inside the upper packer rp may have
slug tests in more complex hydrogeological condi- different values. The riser pipe extends from the upper
tions were presented by Butler et al. (1994). packer to the ground surface and houses the pressure
Hinsby et al. (1992) and Bjerg et al. (1992) pre- transducer and cables. In a multi-level slug test, the
sented an alternative technique for multi-level slug well screen is typically wider than the riser, and
tests which can be efficient in shallow aquifers. This the riser is wider than the packer pipe: rs --> r~ -> rp.
technique employed small-diameter (2.5 cm) driven Lengths of the screen section, riser, and access pipes
wells with short screens (0.25 m). The screen drive are l~, lr, and lp respectively. The depth to the bottom
point was driven in stages to several discrete depths. of the tested well screen section from the static water
The slug test data collected at these depths yielded level is L: L -- l~ + ls +/p. The unconsolidated aquifer
vertical profiles of the hydraulic conductivity. All material is assumed to be locally uniform with respect
the slug test responses were non-oscillatory and
were interpreted using the Dax (1987) method double
which was essentially a simplified Cooper et al. packer z
slug test % ~,
(1967) method. apparatus ~4;~
Multi-level slug tests in highly conductive forma-
tions can be oscillatory. This type of response was
not previously investigated. The hydraulics of a water table T i l ~ N I'.'."
°: rr
double-packer system with components of non- Zone 5-6
uniform diameters differs significantly from the
section l !
hydraulics of the slug test in uniform diameter wells . . . . . . . . . . . . . . . . . . . .

Zone 4-5
used by Springer and Gelhar (1991) or McElwee et al. upper . . . . . . . . .

(1994), and the existing methods and theory must be packer [ "~H El.
extended. Zone 3-4
The purpose of this article is: (1) to present a model s t,oo . . . . .

. . . . . . . . . -:! ~ - - - - ~.---_-: 23 Zone 2-3


for both monotonic and oscillatory responses in a
well-aquifer system of general configuration; and tested
T/
| :.'~ ..t:. ;~'~'; r.
screened ! : ~ r, Zone 1-2
(2) to provide criteria to indicate the type of well section " i
response for different slug test conditions. The accom-
panying article (Zlotnik and McGuire, 1998) will
well screen
I
~U ~:
i:
present: the field procedures involved in a double-
gravel p a c k ~ I [.~
packer slug test; an identification algorithm for K~-
estimation from massive field data; a comparison
between slug test results and grain size analysis of Fig. 1. Schematic diagram of a double-packer system for multi-
disturbed samples recovered at the experimental level slug tests in water-table aquifers.
274 V.A. Zlotnik, ILL. McGuire/Journal of HydroloKv 204 (1998) 271-282

to horizontal hydraulic conductivity K r and vertical (2) the flow through the well screen is uniformly dis-
hydraulic conductivity Kz. The water is assumed to tributed, (3) the change in momentum from the water
be of constant density. It is assumed that the water velocity changing from radial flow through the well
table is the upper horizontal boundary o f the aquifer, screen to vertical flow in the well bore can be
and the lower boundary is impermeable. The com- neglected, and (4) water is incompressible (Cooper
pressibility of the aquifer is neglected following the et al., 1965; van der Kamp, 1976; Kipp, 1985).
Springer and Gelhar (1991) and McElwee et al. (1993) The momentum balance equation for zone 1-2 is
models.
The general model consists of three submodels: d ~V2ls) = -pv~ +p, -p2-ogls-~ (1)
well hydraulics, well-aquifer interaction, and mass
balance. where 0 is water density, V2 is the velocity at eleva-
tion 2 near the entry to the transitional zone 2 - 3 , P l
2.1. Well hydraulics and P2 are the values of the pressure at elevations
1 and 2, respectively, g is acceleration caused by
2.1.1. Momentum equations for pipes in a double- gravity, and fs is the friction losses term. The factor
packer system of ½ is a consequence of a linear change in water
The basic aspects for flow in uniform diameter velocity from the bottom of zone 1-2 to the top
pipes were covered by Cooper et al. (1965), van der (assumption (3)).
Kamp (1976), Springer and Gelhar (1991), McElwee Applying momentum balance to zone 3 - 4 one
et al. (1993), Stone and Clarke (1993). This model can obtains the equation
be extended to a double-packer system where head
d (P V3lp) =P3 - P 4 - pglp - fp (2)
losses occur near contractions and expansions at
pipe connections by applying momentum analysis. where V3 is the uniform velocity inside the packer, P3
Kipp (1985) developed a model for a single pipe and P4 are the pressures at elevations 3 and 4, respec-
contraction or expansion; however this model is not tively, andfv is a friction losses term.
suited for double-packer systems with multiple The momentum balance for water within zone 5 - 6 ,
contractions and expansions. considering water table movement, results in the
For derivation purposes, it is convenient to divide equation
the tested well into five zones between six character-
istic elevations z~ from the bottom of the tested inter- d[o Vs(lr + w)] =P5 + O - og(lr + w) - f (3)
val (Fig. 1). An index i = 1,...6, is used to identify
physical parameters at various elevations. Zone 1 - 2 where w is the water level deviation from the static
is between the lower and upper packers; it is the iso- potentiometric surface (water table), positive upward,
lated screened well section o f length ls. Zone 2 3 P5 is pressure at elevation 5, V5 is the uniform water
represents the short transitional flow zone between velocity inside the riser pipe, andfp is a friction losses
the screened well section and pipe inside the upper term. Pressure at the water table is assumed to be
packer; this zone is a few well diameters long zero.
(Munson et al., 1990). Zone 3 - 4 is inside the access
pipe o f the upper packer which extends from the 2.1.2. Minor losses
screened well section to the riser pipe for the length Minor losses from sudden contractions and expan-
l v. Zone 4 - 5 represents a short transitional flow zone sions o f pipe diameter in a double-packer system for
between the access pipe and the riser pipe. Zone 5 - 6 oscillatory flow processes can be estimated using
is the distance between the top of the transitional zone experimental data (e.g. Munson et al. (1990)). When
4 - 5 and the static water table or length lr. More zones flow is downward, the transitional zone 4 - 5 is con-
can be used for more complex packer systems. sidered as a contraction (Fig. 2a) and zone 2 - 3 is an
The assumptions for flow momentum analysis are enlargement (Fig. 2b). When flow is upward, zone
as follows: (1) the average water velocity in the well is 4 - 5 is considered as an enlargement (Fig. 2c) and
approximately constant over the pipe cross-section, zone 2 - 3 is a contraction (Fig. 2d).
V.A. Zlotnik, V.L. McGuire/Journalof Hydrology 204 (1998) 271 282 275

(a) ~ V (c) ~V

~ ~ r~__~ riser ~ p riser 5


pipe
.Plpe 4
_r~ access ess
pipe e

(b) ,~V (d) ,~ V

access access
pipe pipe
3
I I
rs ~1 tested pl tested
well well
2 I section I section
|. . . .
I
Fig. 2. Pipe hydraulics at locations of sudden changes in pipe diameter in a double-packer system near the access pipe. (a) Contraction at the
water entry to the access pipe (downward flow in zone 4-5). (b) Enlargement at the water entry to the tested screen section in the zone 2 3
(downward flow). (c) Enlargement at the water exit from access pipe in zone 4 5 (upward flow). (d) Contraction at the water entry from the
tested section in zone 2 3 (upward flow).

For sudden pipe contraction or expansion, the where k~3 is the contraction loss coefficient.
difference between head at the entry hen and head Head loss during downward flow in transitional
after exit h~x from a transitional zone is expressed as zone 2 - 3 is
V2 P/$2.~ /°~z2~ e /Og2 (6)
hen-hex =k V2 h = P - z + -- (4) ( P 2 + ~ 2Y--(P3+ 5 3 ) = - k 2 3 ~ 3
2g' og 2
where k~3 is the enlargement loss coefficient. The
where local head h is expressed through pressure p,
general form for transitional zone 2 - 3 valid for
elevation z, and water velocity V at a given position
both flow directions is
(the entry or exit near contraction or enlargement).
Vm is minimal velocity between entry and exit velo- (P3 + P
~ ~2,
3 !-- (P2 + 0~ ~2,
2 ! = -- k23 2V2sgn(V3) (7)
cities, k is the empirical loss coefficient which
depends on the Reynolds number, Re=2Vr/v, and where sgn(x) is a sign function. A generalized minor
the change in pipe area, A = A s/Al, where r is the head loss coefficient, k23, which is valid for both
radius of any given pipe, v is the kinematic viscosity upward and downward flow through sudden changes
of water (v = 1.3 x 10 -6 m 2 s -~ for 10°C), As is the in pipe diameter is
area of the smaller pipe, and A~ is the area of the
{ k~3=kc(A23)' V3>O (rsP)2
larger pipe. k values are available from experiments k23 = e A23 ' (8)
for a broad range of Re (e.g. Munson et al. (1990)). In k23 =ke(A23), V3 < 0
highly developed turbulence regimes (Re > 2000), where empirical functions kC(A) and U(A) are
one may neglect the variability in k on Re. available from experimental data (e.g., Munson
Head loss during upward flow in transitional zone et al. (1990), Figure 8.30 and Figure 8.32, respec-
2 - 3 , assuming its short length (z2 ~ z3) is as follows tively).
Pv2~ P 2 P~Z2 Applying the same concepts at transitional zone
(/93+ ~ 31--(/O2 + ~ V ~ ) = - k ~ 3 ~ 3 (5)
4 - 5 , one obtains a general equation for minor
276 V.A. Zlotnik, V.L. McGuire/Journal of Hydrology 204 (1998) 271 282

head losses well that considers turbulent, laminar, and transi-


PV2"~ In + PV2"~ P 2 tional regimes in the double-packer system, and
(p5-{-~ 51--t/a 4 ~ 4)=-k45~V~sgn(V4) (9) roughness of the slotted well screen, would compli-
cate the analyses (Kaleris, 1989; McElwee et al.,
where
1994). The practical need for these refinements
may be reduced by using small initial water level
k45 = A45 = (l 0) displacements which minimizes the water velocity
k~s =kC(A45), Vs < 0' 7rr
in the double-packer system. Additional studies on
Commonly, values of k23 and k4s which depend on the effect of the initial displacements on slug
the double-packer system configuration are less test results need to be conducted (McElwee et al.,
than 1. 1994).

2.1.3. Friction losses 2.1.4. Hydraulics equation .for the double-packer


Previous studies have addressed friction losses system
during slug testing. Cooper et al. (1965) and van der The hydraulics equation for the double-packer
Kamp (1976) concluded that friction losses can be system can be obtained by combining the momentum
safely neglected after quantitative analysis of both balance equations for all zones (Eqs. (1)-(3), (7) and
laminar and turbulent flow regimes in slug tests. (9)) and using Eq. (11)
This result follows from the analysis of the friction
factor in the Darcy-Weisbach formula for head losses 2 V2 + lpg3+(lr +w)Vs] =Pl-g(L+w)o
~dt 1[ls
in a cylindrical pipe for a laminar regime (~. = 64~Re,
Re < 2000) and a turbulent regime (Blasius formula, 1
+ ~(V2+V2-NR3V3[V3[-N45V4[V4 l) (12)
~ 0.316~Re ¼,2000 < Re < 105) (e.g. Munson et al.,
1990). Using mass continuity of flow in pipes
The role of this term can also be investigated using
numerical experiments. Kabala et al. (1985) studied a //3=I/4, V2r2= V3r~= Vsr2 (13)
mathematical model for a slug test which contained and kinematic considerations
a quadratic term equivalent to the friction losses
term during turbulent flow. The results indicate that dw
V5 = d t ' sgn( V3) = sgn( V4) = sgn( Vs) (14)
this term is insignificant, and a linearized equation
can be used to describe the well response. Stone and one obtains an equation for water level elevation
Clarke (1993) performed numerical experiments con- about the static water level w(t)
sidering turbulence both in the well and the aquifer.
[ dw] (~)(dw) 2
Again, the simulations results were also relatively d (Le+W) d t =Pl
p -g(L +w)- -1
insensitive to friction losses.
In highly permeable aquifers, the magnitudes of (15)
water displacement and velocity decay rapidly during where Le is the effective length of the water column
oscillations around the static water level. Therefore,
2 2
one should expect that friction losses would also ls rr " rr " (16)
L e = -7+ lp . ~ + l r
decay rapidly and make a simplified hydraulic 2 -s rp
model that neglects friction losses more accurate as
and ~"is the coefficient of minor losses
the slug test progresses in time.
Applying this reasoning to the hydraulic model for rr
44r ()
dw
the double-packer system, one can neglect friction ~"= 1 - -rs-4 + .~4
rp k23 + k 4 5 ) s g n ~-
losses in zones 1-2, 3-4, and 5-6
Pressure p i can be replaced with hydraulic head at the
fs ~ 0;fp ~ 0 ; f ~ 0 (11)
screen (inside the well or at the screen surface facing
A more detailed hydraulic model of the flow in the the aquifer). Since z is positive upward, the head at
KA. Zlotnik, KL. McGuire/Journalof Hydrology 204 (1998) 271 282 277

the bottom of the screen is hydrogeological test conditions as follows

h~=P' -ls (181


Pg (22)
and it can be assumed that the hydraulic head in P:P lTVVzz '
different screen points is independent of elevation
(e.g. Kipp, 1985, p.1399). Both the Bouwer and Rice (1976) and Hvorslev
The static hydraulic head in the well and aquifer is (1951) models can be converted to this form (Zlotnik,
1994). It is assumed that K~/Kz is available from
h 0 = lr + lp (19) independent tests; otherwise isotropic conditions are
assumed, Kr = Kz = K.
because the origin of the coordinates is at the top of
the screen. The momentum equation for the whole
double-packer system following from Eqs. (15) and 3. Modified Springer-Gelhar model of the
(18) and Eq. (19) can be given as follows multi-level slug test

, , _ _(dw 2 3.1. Initial value problem


"Le+W" d t e - g ' h s - h ° ' + g w = 2\ dtJ (20)

Combining Eq. (20) and Eq. (21), and considering


the conservation of mass in the well
2.2. Aquifer hydraulics 2 dw (23)
Q=Trrr d-T
Various studies indicate a remarkable validity
one obtains
of the Bouwer and Rice (1976) or Hvorslev (1951)
models for monotonic well responses of water level
in a wide range of parameters (see Widdowson et al. ( L e + w ) ~ -+ 2Krls~ dt + g w = - - , / ~ - (24)
(1990), Hyder and Butler (1995), Brown et al.
where Kr is the only aquifer parameter, since Le is
(1995)). This type of model assumes that the diffusiv-
given by Eq. (16), and estimates of shape factor P for
ity of the aquifer is so high that water changes in the
different test geometries are available from Hvorslev
well are instantaneously reflected in the aquifer. These
(1951), Bouwer and Rice (1976), Dagan (1978),
quasi-steady-state flow models neglect aquifer com-
Widdowson et al. (1990), and Zlotnik (1994).
pressibility, which is often an adequate assumption
Initial conditions describe a falling head test
for relatively small-scale slug tests.
(positive initial displacement w0, where w0 > 0) or
Oscillatory slug test responses were described by
rising head test (negative initial displacement -w0)
Springer and Gelhar (1991) using the Bouwer and
depending on the technique of the slug test. In
Rice (1976) model and by McElwee et al. (1993)
addition, the initial water velocity is zero. These
using the Hvorslev (1951) model for short-screen
conditions can be formulated as
wells. Both models follow the approach of Dagan
(1978) who proposed a simple expression for all w(0)= + w0; d w ( 0 ) - 0 , w 0 > 0 (25)
quasi-steady state models dt
Q The initial value problem Eqs. (24) and (25) is valid
h0 - hs = - - (21) for slug tests in wells with non-uniform pipe
2 7rKrP l s
diameters using a double-packer system.
where Q is discharge to the well, and P is a shape By omitting the minor losses term and considering
factor. P depends on the test geometry, the vertical wells of uniform diameter (rs -- rp rr) , one obtains
=

anisotropy of the aquifer, the aquifer saturated thick- the Springer and Gelhar (1991) model. Note that in
ness b, the "well radius" rw which accounts for this model, the parameter Le = ls/2 + lp + lr indicates
gravel pack around the borehole (if present) and the the length of the effective water column (from the
278 ILA. Zlotnik, V.L. McGuire/Journal of Hydrology 204 (1998) 271-282

static water table to the center of the tested well and well parameters
section). l
Since this analysis of well hydraulics differs from F(y)=FoH(v), H(v)=
1 +ywo/L e' G(y,y')
the Springer and Gelhar (1991) work by the coeffi-
r4
cient Le and the minor losses term, our model Eqs. + rr
= ~(v')H(v), ¢(v') = 1 - k23 + k45)sgn(y');
.-~(4
(24) and (25) will be referred to as the modified r4 rp
Springer-Gelhar (SG) model. In a practical sense,
the minor losses term is not very significant, because Fo= r2 (g---) I/2
(28)
coefficient ~"values are on the order of 0.1-1.0 (e.g. 2KrPls \ L J
Munson et al. (1990)). This model lends itself to straightforward lineariza-
Minor losses can also be neglected in the SG model tion and analysis. Assuming small drawdown pertur-
if approximate K-estimates are acceptable (Bouwer, bations (wo/Le << 1) and omitting the second order
1996). Then only the shape factor P is required for terms in y and y', one obtains a linear Eq. (27) with
data input similar to the Bouwer and Rice (1976) coefficients
method• However, one can retain the minor losses
term to improve the match between field data and F(y) ~ Fo, H(V) --~ 1, G(y,y') ~ 0 (29)
simulated well responses.
One should note that the SG model is limited which was used by Springer and Gelhar (1991). This
to highly conductive aquifers and it cannot replace linearization is valid when perturbations of water
the Cooper et al. (1965), van der Kamp (1976), level are small compared with the effective length
Krauss (1977), or Kipp (1985) models where of the water column. The linearization is invalid
aquifer compressibility plays an important role in when the short tested well screen is too close to the
aquifer hydraulics. The SG model's simplicity is water table.
caused by the quasi-steady-state model for well The solution of this equation is (Springer and
aquifer interaction, and computations can use efficient Gelhar, 1991):
numerical methods (e.g., the Runge-Kutta method
1. overdamped (monotonic) if F0 > 2
(Press et al., 1989)). This makes the SG model very
effective for massive field data interpretations. _+1
y(r)= (7_e~+~-3~+e v r),
3/_ -5'+
3.2. Linear analysis of slug test model 7+_ = - Fo/2 + (F2/4 - 1) 1/2 (30)

Using dimensionless variables and another notation


2. critically damped (transition between monotonic
for the time derivative
and oscillatory responses), if F0 = 2
y(r)= -+ e-~(1 +r) (31)
I'g'~ l/z w , dy
r=l ~Le) ' y= -W- 0' Y =-~T
3. underdamped (oscillatory) if F0 < 2

the generalized SG model is as follows y(r)= + e F"r/2Icos('yT)+ ~--~°sin(y'r)],q/ J

y"=F(y)y' +H(y)y= - G(y,y')(y) 2, y(0)= + 1, 7=(1 - F 2 / 4 ) 1/2 (32)

y'(0)=0 (27)
Generally, the non-linear initial value problem
where the plus sign is used for the falling head test, Eqs. (24) and (25) can be solved using numeri-
and the minus sign is used for the rising head test, and cal methods for slug test simulation or estimation
the coefficients F, H and G depend on both aquifer of Kr.
ILA. Zlotnik, ILL. McGuire/Journal of Hydrology 204 (1998) 271-282 279

3.3. Correspondence between Springer-Gelhar and from the water table to the bottom of tested screen
B o u w e r - Rice models section is 5.24 m (which is slightly smaller than L =
ls + lr + lp = 5.37 m because o f packer design).
The SG model predicts that multi-level slug tests in For all examples, assume the aquifer has a vertical
the same well at the same elevation can exhibit both anisotropy (Kr/Kz) in a range o f 1 to 10, and a pre-
monotonic and oscillatory behavior depending on the dicted K r range from 10 to 200 m/day. Our analysis
test setup. allows us to predict types o f test responses in this Kr
The simple criterion F0 < 2 can be used to deter- range. We will proceed with analysis of anisotropy
mine when oscillatory response is expected from the effects on the potential o f oscillatory slug test
test. The response will be oscillatory when aquifer responses. The linear model Eqs. (29)-(32) requires
l o c a l K r is larger than a critical value Kc which estimates o f effective length Le and shape factor P.
depends on both the double-packer system, slug test Effective length is estimated from Eq. (16): Le =
geometry, and aquifer characteristics 5.4 m. The shape factor can be obtained from either
Bouwer and Rice (1976) method or Widdowson et al.
(1990) data (since b/ls = 18.5 which is approximately
K~ > K c = 4l~P\ L e j
(33) in the range o f validity o f the latter).
In the first example, assuming isotropic conditions,
At the same elevation, the double-packer system P = 0.40 (Widdowson et al., 1990, Figure 7) and Kc =
parameters (position and length of the tested interval, 73 m/day. This means that a slug test response will
effective screen length, and riser pipe radius) control be oscillatory if local K at this elevation is larger
the type o f response. than 73 m/day.
If F0 > 2, monotonic response is expected. Tradi- In the second example, assuming anisotropic con-
tionally, a monotonic response is analyzed by the ditions (Kr/Kz = 10), P = 0.32 from Widdowson et al.
Bouwer and Rice (1976) method. The model is (1990), Table 5) or from the Bouwer and Rice (1976)
equivalent to the Bouwer and Rice (1976) model if method corrected for aquifer anisotropy (Ztotnik,
F0 >>2. In this case, y+ ~ - F o , y -~ - 1/Foand 1994), and Kc = 91 m/day. This means that a slug
Eq. (30) becomes: y(r) -~ exp(-r/Fo). This is exactly test response will be oscillatory if local Kr is larger
the Bouwer and Rice (1976) equation (see also than 91 m/day.
(Zlotnik and Ferlin, 1994)) The third example illustrates the effect o f changing
the double-packer system design for isotropic condi-
w(t) = woe at, X= 2 KrlsP tions. The riser pipe radius is reduced from r~ =
r2 (34) 0.026 m to rr = 0.013 m. The effective length, shape
factor, and critical value of the hydraulic conductivity
where P = ln(Re/rw) is the Bouwer and Rice (1976) for a given elevation are Le = 5.4 m, P = 0.40, and Kc
shape factor and Re is the radius o f influence. The = 73 m/day, respectively, for the riser pipe radius rr =
shape factor can also be estimated using Widdowson 0.026 m. When the riser pipe radius is decreased rr =
et al. (1990) data. 0.013 m, these values become Le = 4.3 m, P = 0.40,
and Kc = 20 m/day. This indicates that a slug test
response for a double-packer system with a narrow
4. Discussion riser pipe will be oscillatory if K~ = 44 m/day, and
monotonic if the double-packer system has a larger
The developed concepts can be illustrated by riser pipe.
examples o f double-packer systems employed in our The simulated slug test responses are shown on
studies (Zlotnik and McGuire, 1998) with the follow- Fig. 3 for an isotropic aquifer with K r = 44 m/day.
ing set of parameters (Fig. 1): ls = 0.67 m, lr = 3.9 m , lp First, slug test responses with a larger diameter
= 0.8 m, inside screen radius rs = 0.051 m, outside riser pipe radius are computed using the linear
well radius rw = 0.057 m, rr = 0.26 m, and rp = model Eq. (30). This model predicts monotonic
0.019 m, aquifer thickness b = 12.4 m, and depth response because F0 = 3.65 (Fig. 3a) which can be
280 V.A. Zlotnik, V.L. McGuire/Journal of Hydrology 204 (1998) 271-282

0.0

-0.1

-0.2

-0.3
-0.4
o -0.5

-0.6
-0.7
~ '/ - - Ilnearlzed SG model (Eq. 35)
/l . . . . Bouwer and Rice model (Eq. 36)
.... non-linear SG model (Eq. 27)
-0.6

-0.9
-1.0 i i i [ i i i

0 2 4 6 8 10 12 14
t(seconds)

0.4

0.2

0.0

-0.2

-0.4
• Ilnearlzed SG model (Eq. 37)
-0.6

-0.8

I I i i i l i
-1.0
2 4 6 8 10 12 14
I (seconds)
Fig. 3. Multi-level slug test responses for different double-packer system configurations. (a) Monotonic response, (b) oscillatory response.

expressed as follows The slug test response with a smaller riser pipe
radius was computed using the linear model Eq. (32)
W/Wo= - 1.098e-°4°2t + 0.098e 4.szt (35)

where t is given in seconds. W/Wo= - e - 0 '654t [cos(1.31t)+O.577sin(1.31t)] (37)


Second, the same slug test response is simulated
using the Bouwer and Rice (1976) formula Eq. (34) which exhibits oscillatory response because F0 = 1.0
and 3~ = 0.866 (Fig. 3b).
w/wo = - e-°'367t (36)
In practice, one can predict the type o f slug test
which predicts a slightly slower water level recovery response using criterion Eq. (33) and the preliminary
(Fig. 3a). The parameter ~"for the non-linear model is evaluations o f K from analysis o f drilling logs or other
estimated using Eqs. (28) and (8) and Eq. (10), where data on aquifer lithology.
A23 =0.14, A45 =0.53, kC(Az3) = 0.5, kC(A45) = 0.25, It is important to mention that some non-linear
ke(A23) = 0.7, ke(A45) = 0.21, (Munson et al., 1990 physical processes which are involved in the slug
Figure 8.30 and Figure 8.32). The results o f linear- test are simplified in this model. Therefore one
ized and non-linear models are similar. should not overestimate the rigor o f the SG model;
V.A. Zlotnik, V.L. McGuire/Journal of Hydrolog), 204 (1998) 271 282 281

this m o d e l is r e c o m m e n d e d o n l y in highly p e r m e a b l e References


aquifers.
Belitz, K., Dripps, W., Fairchild, J.E., 1995. Slug tests in elastic,
unconfined aquifers: incorporating the free surface, Water
Resources Res., 31 (12), 3329-3336.
5. Conclusions
Bjerg, P.L., Hinsby, K., Christensen, T., Gravesen, P., 1992. Spatial
variability of hydraulic conductivity of an unconfined sandy
The S G m o d e l for m u l t i - l e v e l slug tests in h i g h l y aquifer determined by a mini slug test, J. Hydrol., 136, 107
p e r m e a b l e aquifers c o m b i n e s features o f the Springer 122.
and G e l h a r (1991) and B o u w e r and R i c e (1976) Bouwer, H., 1996. Discussion of Bouwer and Rice slug test review
articles, Ground Water, 34 (1), 171.
models. This m o d e l is relatively simple and simula-
Bouwer, H., Rice, R.C., 1976. A slug test for determining
tions can be p e r f o r m e d using standard n u m e r i c a l hydraulic conductivity of unconfined aquifers with completely
m e t h o d s w h i c h m a k e s it efficient for m a s s i v e field or partially penetrating wells, Water Resour. Res., 12 (3),
data interpretations, W h e n slug tests are p e r f o r m e d 423 -428.
without a d o u b l e - p a c k e r system in a u n i f o r m d i a m e t e r Bredehoeft, J., 1967. Response of well-aquifer systems to Earth
tides, J. Geophys. Res., 72 (12), 3075-3087.
well, the S G m o d e l reduces to the Springer and G e l h a r
Brown, D.L., Narasimhan, T.N., Demir, Z., 1995. An evaluation of
(1991) model. W h e n slug tests are p e r f o r m e d in a the Bouwer and Rice method of slug test analysis, Water
l o w e r c o n d u c t i v i t y aquifer, the S G m o d e l reduces to Resour. Res., 31 (5), 1239 1246.
the B o u w e r and R i c e (1976) model. Butler, J.J., Bohling, G.C., Hyder, Z., McElwee, C.D., 1994. The
The S G m o d e l indicates that m u l t i - l e v e l slug tests use of slug tests to describe vertical variations in hydraulic
conductivity, J. Hydrol., 156, 137 162.
in the s a m e well at the s a m e e l e v a t i o n can exhibit
Butler, J.J. Jr., McElwee, C.D., Liu, W., 1996. Improving the qual-
m o n o t o n i c or oscillatory w a t e r level response, a situa- ity of parameter estimates obtained from slug tests, Ground
tion not p r e v i o u s l y described in the literature. The Water, 34 (3), 480-490.
type o f response for a d o u b l e - p a c k e r system depends Cooper, H.H. Jr., Bredehoeft, J.D., Papadopulos, S.S., 1967.
on a c o m b i n a t i o n o f several factors: aquifer hydraulic Response of a finite diameter-well to an instantaneous charge
of water, Water Resour. Res., 3 (1), 263-269.
conductivity, screen length o f the tested interval, riser
Cooper, H.H. Jr., Bredehoeft, J.D., Papadopulos, I.S., Bennett,
pipe radius, e f f e c t i v e length o f the w a t e r column, and R.R., 1965. The response of well-aquifer system to seismic
other parameters e m b e d d e d in the B o u w e r and R i c e waves, J. Geophys. Res., 70 (16), 3915 3926.
(1976) or W i d d o w s o n et al. (1990) shape factor. The Dagan, G., 1978. A note on packer, slug, and recovery tests in
relation b e t w e e n hydraulic c o n d u c t i v i t y and slug test unconfined aquifers, Water Resour. Res., 14 (5), 929 934.
Dax, A., 1987. A note on the analysis of slug tests, J. Hydrology,
parameters allows us to estimate the type o f slug
91, 153-177.
test response. In practice, one can estimate the type Hess, K.M., Wolf, S.H., Celia, M.A., 1992. Large scale natural
o f well response based on analysis o f drilling logs or gradient tracer test in sand and gravel, Cape Code, Massachu-
lithology. setts. 3. Hydraulic conductivity variability and calculation
macrodispersivities, Water Resour. Res., 28 (8), 2011-2027.
Hinsby, K., Bjerg, P.L., Andersen, L.J., Skov, B., Clausen, E.V.,
1992. A mini slug test method for determination of a local
Acknowledgements hydraulic conductivity of an unconfined sandy aquifer, J.
Hydrol., 136, 87-106.
This w o r k was supported in part by the W a t e r Hvorslev, M.J., 1951, Time Lag and Soil Permeability in Ground-
C e n t e r o f the U n i v e r s i t y o f Nebraska; the U S G S water Observations, U.S. Army Corps of Eng., Waterways Exp.
annual federal m a t c h i n g grant as authorized by Stn. Vicksburg, MS. Bull. 36, 56 pp.
Hyder, Z., Butler, J.J. Jr., 1995. Slug tests in unconfined
Section 104 o f the W a t e r R e s e a r c h Act; and the Cen-
formations: an assessment of the Bouwer and Rice technique,
tral Platte Natural R e s o u r c e s District, Nebraska. W e Ground Water, 33 (1), 16-22.
are grateful to L.W. G e l h a r ( M I T ) for p r o v i d i n g us Hyder, Z., Butler, J.J., McElwee, C.D., Liu, W., 1994. Slug tests in
with information on the slug test method. W e are partially penetrating wells, Water Resour. Res., 30 (11), 2945
also indebted to C.D. M c E l w e e ( K G S ) and J.J. Butler, 2957.
Kabala, Z., Pinder, G.F., Milly, P.C.D., 1985. Analysis of well-
Jr. ( K G S ) for r e v i e w i n g an early v e r s i o n o f this article,
aquifer response to a slug test, Water Resour. Res., 21 (9),
and G. van der K a m p ( N H R I , Canada) for fruitful 1433-1436.
discussions. Kaleris, V.I 1989, Inflow into monitoring wells with long screens.
282 EA. Zlotnik, V.L. McGuire/Journal of Hydrology 204 (1998) 271-282

In: H.E. Kobus and W. Kinzelbach (Eds.), Contaminant Trans- Rojstaczer, S., 1988. Determination of fluid flow properties from
port in Groundwater, A.A. Balkema, Rotterdam, 41-50. the response of water levels in wells to atmospheric loading,
Kipp, K.L. Jr., 1985. Type curve analysis of inertial effects in the Water Resour. Res., 24 (11), 1927 1938.
response of a well to a slug test, Water Resour. Res., 21 (9), Shapiro, A.M., 1989. Interpretation of oscillatory water levels in
1397-1408. observation wells during aquifer tests in fractured rock, Water
Krauss, 1., 1977, Determination of the transmissibility from the Resour. Res., 25 (10), 2129-2137.
free water level oscillation in well-aquifer systems. In: Springer, R. K., L. W. Gelhar, 1991. Characterization of large
H.J. Morel-Seytoux (Ed.), Surface and Subsurface Hydrology, scale aquifer heterogeneity in glacial outwash by analysis of
Proceedings of the Fort Collins 3rd International Hydrology slug tests with oscillatory response, Cape Cod, Massachusetts.
Symposium, Water Resources Publications, Fort Collins, In: G. Mallard and D. Aronson (Eds.), U.S. Geological Survey,
Colo., 268-279. Toxic Substances Hydrology Program, Proceeding of the Tech-
McElwee, C.D., Butler, J.J., Jr., et al., 1993. Characterization of nical Meeting, Monterey, California, March 11-15, 1991,
heterogeneities controlling transport and fate of pollutants in USGS, WRI Rept. 91-4034, Reston, Virginia, 36-40.
unconsolidated sand and gravel aquifers: Second year report, Stone, D.B., Clarke, G.K.C., 1993. Estimation of subglacial
Publication Office, Kansas Geological Survey, Report KGS hydraulic properties from induced changes in basal water
OFR 93-21. pressure: a theoretical framework for borehole response tests,
McElwee, C.D., Butler, J.J., Jr., et al., 1994. Characterization of J. of Glaciology, 39 (132), 327-340.
heterogeneities controlling transport and fate of pollutants in van der Kamp, G., 1976. Determining aquifer transmissivity by
unconsolidated sand and gravel aquifers: Third year report, means of well response tests: the underdamped case, Water
Publication Office, Kansas Geological Survey, Report KGS Resour. Res., 12 (1), 71 77.
OFR 94-32. Widdowson, M.A., Molz, F.J., Melville, J.G., 1990. An analysis
Melville, J.G., Molz, F.J., Guven, O., Widdowson, M.A., 1991. technique for multilevel and partially penetrating slug test data,
Multilevel slug tests with comparisons to tracer data, Ground Ground Water, 28 (6), 937-945.
Water, 29 (6), 897-907. Zlotnik, V., 1994. Interpretation of slug and packer tests in aniso-
Munson, B.R., Young, D.F., Okiishi, T.H., 1990, Fundamentals of tropic aquifers, Ground Water, 32 (5), 761 766.
Fluid Mechanics, John Wiley, New York, p. 497-516. Zlotnik, V., M. Ferlin, 1994, Vibracoring technique for well instal-
Neuzil, C.E., 1982. On conducting the modified "slug" test in tight lation and slug testing. In: Proceedings, The Eighth Outdoor
formations, Water Resour. Res., 18 (2), 439-441. Action Conference and Exposition, Aquifer Remediation,
Press, W.H., Flannery, B.P., Teukolsky, S.A., Vetterling, W.T., Ground Water Monitoring, Geophysical Methods, May 23-25,
1989. Numerical Recipes (Fortran Version), Cambridge 1994, NGWA, 647-661.
University Press, Cambridge. Zlotnik, V.A., McGuire, V.L., 1998. Multi-level slug tests in highly
Ritzi, R.W., Sorooshian, S., Gupta, V.K., 1991. On the estimation permeable formations: 2. Hydraulic conductivity identification,
of parameters for frequency domain models, Water Resour. method verification, and field applications., J. Hydrol., 204,
Res., 27 (5), 873 882. 283-296.

You might also like