3 WATER RESOURCES
RESEARCH JUNE 1976
A SlugTestforDetermining
Hydraulic
Conductivity
ofUnconfined
Aquifers
With Completely
or PartiallyPenetrating
Wells
HERMAN BOUWER AND R. C. RICE
U.S.Water
Conservation
Laboratory,
Agricultural
Research
Service,
U.S.Department
ofAgriculture,
Phoenix,
Arizona85040
A procedure
ispresented
forcalculating
thehydraulic
conductivity
ofanaquifer
nearawellfromthe
rateof riseof thewaterlevelin the wellaftera certainvolumeof wateris suddenlyremoved.
The
calculation
is basedontheThiemequation of steady
stateflowto a well.Theeffective
radius
Reover
which
theheaddifference
between
theequilibrium
watertableintheaquifer
andthewaterlevelinthewell
isdissipated
wasevaluated
witha resistance
networkanalog
fora widerange
ofsystemgeometries.
An
empirical
equation
relating
Retothegeometryof thewellandaquiferwasderived.
Thetechnique
is
applicable
tocompletely
orpartiallypenetrating
wellsinunconfined
aquifers.
It canalsobeused
for
confined
aquifers
thatreceive
waterfromtheupper
confining
layer.
Themethod'sresults
arecompatible
withthoseobtained
by othertechniques
for overlapping
geometries.
Iny = --r•2
In(Re/r,o)
+ constant (4)
Applying this equation betweenlimits yo at t = 0 and Yt at t
and solvingfor K yield
K = rc ln(R,/r,o)
2L
1lnYø
t yt
(5)
This equation enablesK to be calculatedfrom the rise of the
waterlevelin the well after suddenlyremovinga slugof water
// / / ///////// / / / ///•
from the well. SinceK, r•, rw, Re, and L in (5) are constants,
IMPERMEABLE
(l/t) In Yo/yt must also be constant.Thus field data should
Fig. 1. Geometry and symbolsof a partially penetrating,partially yielda straightlinewhentheyareplottedasIn yt versust. The
perforated well in unconfinedaquifer with gravel pack or developed term(l/t) In Yo/Ytin (5) is thenobtainedfromthebest-fitting
zone around perforated section. straightline in a plot of In y versust (seethe example).The
valueof In Re/rw is dependenton H, D, L, and rwand canbe
from well centerto originalaquifer(well radiusor radiusof evaluatedfrom the analogresultspresentedin the next section.
casingplusthicknessof gravelenvelopeor developedzone). The transmissibilityT of the aquifer is calculatedby multi-
The termsL, y, Re, and r•oare all expressed in unitsof plying(5) by the thicknessD of the aquifer or
length.The effectiveradiusReis theequivalentradialdistance
overwhichthe headlossy is dissipatedin the flow system.The T = Drc:
In(R.,/r,o)
2L
1In?-9-ø
t yt
(6)
valueof Redepends on thegeometryof theflowsystem,andit
was determined for different values of H, L, D, and r•o(Figure This equationis basedon the assumption that the aquiferis
1) with a resistance
networkanalog,aswill bediscussed in the uniform with depth.
nextsection.Equation(1) is basedon the assumptions that (1) Equations(5) and (6) are dimensionallycorrect.ThusK and
drawdown of the water table around the well is negligible,(2) T are expressedin the same units as the length and time
flow above the water table (in the capillary fringe) can be parametersin the equations.
ignored,(3) headlossesas waterentersthe well (well losses)
are negligible,and (4) the aquiferis homogeneous and iso- EVALUATIONOF Re
tropic.Theseare the usualassumptions in the development of Values of Re, expressedas In Re/rw, were determinedwith
equationsfor pumpedhole techniques[BouwerandJackson, an electricalresistancenetwork analogfor differentvaluesof
1974, and referencestherein]. r•, L, H, and D (Figure 1), usingthe sameassumptionsas
The valueof r•oin (1) representstheradialdistancebetween thosefor (l). An axisymmetricsectorof I tad was simulated
the undisturbedaquiferand the well center.Thusr•oshould by a network of electrical resistors.The vertical distance be-
includegravelenvelopesor 'developed'zonesif theyaremuch tween the nodeswas constant, but the radial distancebetween
morepermeablethan the aquiferitself(Figure 1). nodesincreasedwith increasingdistancefrom the centerline
The rate of rise,dy/dt, of the waterlevelin the well after (Figure 2). This yielded a network with the highestnode
suddenly removing a slugof watercanberelatedto theinflow densitynearthe well, wherethe headlosswasgreatest,and a
Q by the equation decreasingnode densitytoward the outer reachesof the sys-
tem. For a more detailed discussionof graded networksfor
dy/dt = - Q/;rro2 (2)
representingaxisymmetricflow systems,seeLiebmann[1950]
where;rrc2 is the cross-sectional area of the well wherethe and Bouwer[1960].
water levelis rising.The minussignin (2) is introducedbe- The radial extentof the mediumrepresented on the analog
causey decreases as t increases. was more than 60,000 times the largestr• value used in the
The termrcistheinsideradiusof thecasingif thewaterlevel analyses.Thus the radial extent of the analog systemwas
is abovetheperforatedor otherwise openportionof thewell. essentiallyinfinite,as evidencedby the fact that a reductionin
If thewaterlevelis risingin theperforatedsectionof thewell, radial extentby severalnodesdid not havea measurableeffect
allowanceshouldbe made for the porosityoutsidethe well on the observedvalue of
casingif the hydraulicconductivity of thegravelenvelope or The value of Re for an infinitelydeepaquifer(D = co)was
developed zoneis muchhigherthanthatof theaquifer.In that determinedby simulating an impermeableand then an in-
casethe (open)porosityin the permeable zone mustbe in- finitely permeablelayer at a certainvalue of D. If this value of
cludedin the cross-sectional area of the well. For example,if D is takento be sufficientlylarge,the flow in the systemwhen
the radiusof the perforatedcasingis 20 cm and the casingis the layerat D is taken as beingimpermeable is only slightly
BOUWER
ANDRICE:GROUNDWATER
HYDRAULICS 425
ioo
ß
IOO
ß
200
ß
R
ß
300
ß
4OO
ß
•e ß ß ß ß
ß
I
ß
ß ß ß ß ß ß
5OO
ß
ß ß ß ß ß ß
ß
ß ß ß ß
6OO
--•/.////////////////////// IMPERMEABLE
Fig.2. Nodearrangement
(dots)
forresistance
network
analog
andpotential
distribution
(indicated
aspercentages
on
equipotentials)
forsystem
withL/rw= 625,
H/rw= 1000,
andD/r• = 1500.
Thenumbers
ontheleftandatthetopofthe
figurearearbitrarylengthunits(notebreaksin horizontal
scale).
lessthan the flow whenthe layer is taken as beinginfinitely resultsindicatedthat theeffectiveupperlimit of In [(D - H)/
permeable.
Theaverage of thetwoflowscanthenbetakenasa r•] is 6. Thusif D is considered infinityor (D - H)/rw is so
of theflowthatwouldoccurif theaquiferwere largethat In [(D - H)/r•] is greaterthan 6, a valueof 6
goodestimate
on the analogas beinguniformto infinitedepth should still be used for the term In [(D - H)/r•] in (8).
represented
[Bouwer,
1967].Thisaverage flowwasusedto calculateRefor If D = H, thetermIn [(D - H)/r•] in (8) cannotbe used.
Theanalog
results
indicated
thatfor thiscondition,
whichis
wereperformedby simulatinga system thecaseof a fullypenetrating
The analoganalyses well,(8) shouldbemodified
to
with certain values of rw' H, and D. The electrical current
entering
the'well'wasthenmeasured for different
valuesof L,
rangingfrom nearH to near0. This wasrepeatedfor other lnRe/r,,,
= •n-•/-•)
1.1+L•r•)-' (9)
valuesof rw,H, and D. The conditionWhereL = H couldnot
on the analogbecauseit Wouldmeana short whereC is a dimensionless
be simulated parameterthat is a function'of
betweenthe water tableas thesourceand thewell asthe sink. L/r• as shown in Figure 3.
The electricalcurrent flow in the analogwas convertedto Equations
(8) and(9)yieldvalues
ofInRe/rwthatarewithin
volume with(1) foreach 10%of theactualvalueasevaluated
perday'andIn Re/rwwasevaluated byanalogif L > 0.4Hand
combination of rw,H, L, andD Usedin the analog. within 25% if L << H (for example,L = 0.1H).
For a givengeometrydescribed by r•o,H, andD, thecurrent The analoganalyses wereperformedfor wellsthat were
flowQt intothesimulated linearlywith closed
Wellvariedessentially at thebottom.Occasionally,
however, wellswithopen
L and could be describedby the equation bottoms were also simulated.The flow through the bottom
appeared
tobenegligible
forallvalues
ofrwandL used
inthe
Q, = mL + n (7) analyses.
If L isnotmuchgreater
thanr• (forexample,
L/r•
Because
ofthelinearity
between ofthe << 4), thesystemgeometry
Q,andL theresults approaches
that of a piezometer
analyses
could
beextrapolated L = H. The cavity[Bouwer
tothecondition andJackson,
1974],in whichcasethebottom
values
ofrn
in(7)appeared.to
vary
inversely
with InH/r•.The flow
canbesignificant.Equations
(8)and(9)can alsobeused
values
ofnvariedapproximately
linearly
withIn[(D- H)/ to evaluate
InRe/r•if aportion
openpartofthewellisisolated
oftheperforated
withpackers
orotherwise
fortheslugtest.
r•], theslope
A andintercept
B in theserelations
beinga func'
tion of L/rw, This enabledthe derivationof the followingEquipotentialsfor theflowsystem arounda partiallypene-
empirical
equationrelatingIn Re/r• to the geometry
of the trating,
partially
perforated well
in anunconfined aquifer
after
system:
lowering thewaterlevelin thewellareshown in Figure2. The
numbers alongthesymmetry axisandthewatertablerep/'e-
sentarbitrarylengthunits,Thenumbers ontheequipotentials
InR. 1'1 A+BIn
[(O
--H)/r,o
r•'= _ln(H/r,o)]1-'
q- L/r,o (8) indicatethepotentialasa percentage of thetotalheaddiffer-
encebetweenthe watertable(100%)and the openportionof
In this equation,A and B are dimensionless
coefficients
that the well (0%) shownas a dashedline.
are functionsof L/r•, as shown in Figure 3. If D >> H, an The value of Re for the casein Figure 2 is 96.7 lengthunits.
increasein D has no measurableeffecton In Re/rw. The analog As shownin the figure,thiscorresponds approximately to the
426 BOUWER AND RICE: GROUNDWATER HYDRAULICS
14'
' [ '['['l ][][I ' [ ' [']'l [[Ill , I , I,[,[ I[[] I ' ] ' [']'l
12-
A
and
½
I0 - / _'
/
8-
-.
B
6- ß
4- _
85% equipotentialwhen Re is laterally extendedfrom the cen- rapid infiltrationbasinsfor groundwaterrechargewith sewage
ter of the open portion of the well. Thus most of the head loss effluent[Bouwer,1970].The staticwater table was at a depth
in the flow systemoccursin a cylinderwith radiusRe, whichis of3m, D = 80 m, H = 5.5 m, L = 4.56m, rc = 0.076m, and rw
indicativeof the horizontal extentof the portion of the aquifer was taken as 0.12 m to allow for developmentof the aquifer
sampled for K or T. The vertical extent is somewhat greater around the perforated portion of the casing. A Statham
than L, as indicatedby, for example,the 80%equipotentialin PM131TC pressuretransducerwas suspendedabout I m be-
Figure 2. low the static water level in the well (when trade namesand
To estimate the rate of rise of the water level in a well after it companynamesare included,they are for the convenienceof
is suddenlylowered,(5) can be written as the reader and do not imply preferentialendorsementof a
particular product or company over othersby the U.S. De-
2 Re Yo
r, In- In- (10) partment of Agriculture). A solid cylinder with a volume
t - 2KL r•o Yt equivalentto a 0.32-m changein water level in the well was
By takingYt = 0.9y0,(10) reducesto also placed below the water level. When the water level had
returned to equilibrium, the cylinder was quickly removed.
r, 2 Re The transduceroutput, recorded on a Sargent millivolt re-
t.•o%
ß = 0.0527 -• Inrw -- (11) corder, yieldedthe y-t relationshipshownin Figure 4 with y
wheret90% is the timethat it takesfor the waterlevelto rise plottedon a logarithmicscale.The straight-lineportion is the
90%of the distance to theequilibrium level.By assuming a per- valid part of the readings.The actual Y0 value of 0.29 m
meableaquiferwithK = 30m/day,a wellwithre= 0.2m and indicatedby the straightline is closeto the theoreticalvalue of
L = 10m, andIn (Re/rw)= 3, (11) yieldst90%= 1.82s.Thusif 0.32 m calculatedfrom the displacementof the submerged
Y0is taken as 30 cm, it takes 1.8 s for the water level to rise 27 cylinder.
cm, another 1.8 s for the next 2.7 cm (90% of the remaining3 Extendingthe straight line in Figure 4 showsthat for the
cm), and another 1.8 s for the next 0.27 cm, or a total of 5.4 s arbitrarily selected
t valueof 20s,y = 0.0025m.Thus(l/t) in
for a rise of 29.97 cm. Measurementof this fast rise requiresa Yo/Yt= 0.238 s-•. The valueof L/rw = 38, for whichFigure3
sensitiveand accuratetransducerand a fast-responserecorder. yieldsA = 2.6 and B = 0.42. Substitutingthesevaluesinto (8)
The rate of rise can be reduced by allowing groundwater to and usingthe maximumvalue of 6 for In [(D - H)/rw] (since
enter through only a portion of the open sectionof the well, as In [(D - H)/rw] for thewell exceeds 6) yieldIn (Re/rw)= 2.37.
can be accomplishedwith packers. Equation (5) then givesK = 0.00036 m/s = 31 m/day. This
For a moderatelypermeableaquifer with, for example,K = value agreeswith K values of 10 and 53 m/day obtained
I m/day, a well with rc = 0.1 m and L = 20 m, and In (Re/r•o) previouslywith the tube method on two nearby observation
= 5, (11) yieldst = 11.4 s. In this case,it would take the water wells [Bouwer, 1970]. These K valueswere essentiallypoint
level 22.8 s to rise from 30 cm to 0.3 cm below static level. measurementson the aquifer immediately around the well
bottoms,whichwereat depthsof 9.1 and 6.1 m, respectively.
EXAMPLE
COMPARISONS
A slug test was performed on a casedwell in the alluvial
depositsof the Salt River bed west of Phoenix, Arizona. The œiezometer method. The geometryto which(8) and (9) and
well, known as the east well, is located about 20 m east of six the coefficients
in Figure 3 apply overlapsthe geometryof the
BOUWER
ANDRICE:GROUNDWATER
HYDRAULICS 427
K - •rr,o'"
Art
1Iny__•o
Yt
(12)
whereAr is a geometryfactor with dimensionof length. Val-
uesof A r were evaluatedwith an electrolytictank analogby
Youngs[1968], whoseresultswere expressedin tabular form as
A•,/r•, for differentvaluesof L/r•, (rangingbetween0 and 8),
(H - L)/r,,,, and (D - H)/r,,,.
Taking a hypotheticalcasewhereLira, = 8, Hire, = 12, and
D/r,,, = 16, K calculatedwith (5) is 18% below K calculated
with (12). This is more than the 10%error normallyexpected
with (8) and (9) for the L/H value of 0.67 in this case.The
larger discrepancymay be due to the differencein method-
ology, or to the fact that the L/r,,, value is closeto the lower
limit of the rangecoveredon the resistancenetwork analog.
An approximateequation for calculatingK with the pie- 0.01