WATER RESOURCES RESEARCH. VOL,. 20. NO. 7, PAGES 831-846.

J U L Y I984

Double-Porosity Models for a Fissured Groundwater Reservoir With Fracture Skin

U . S . Geological Survey, Menlo Park, Calgornia

Theories of flow to a well in a double-porosity groundwater reservoir are modified to incorporate effects of a thin layer of low-permeability material or fracture skin that may be present at fracture-block interfaces as a result of mineral deposition or alteration. The commonly used theory for flow in doubleporosity formations that is based upon the assumption of pseudo-steady state block-to-fissure flow is shown to be a special case of the theory presented in this paper. The latter is based on the assumption of transient block-to-fissure flow with fracture skin. Under conditions where fracture skin has a hydraulic conductivity that is less than that of the matrix rock, it may be assumed to impede the interchange of fluid between the fissures and blocks. Resistance to flow at fracture-block interfaces tends to reduce spatial variation of hydraulic head gradients within the blocks. This provides theoretical justification for neglecting the divergence of flow in the blocks as required by the pseudo-steady state flow model. Coupled boundary value problems for flow to a well discharging at a constant rate were solved in the Laplace domain. Both slab-shaped and sphere-shaped blocks were considered, as were effects of well bore storage and well bore skin. Results obtained by numerical inversion were used to construct dimensionless-type curves that were applied to well test data, for a pumped well and for an observation well, from the fractured volcanic rock terrane of the Nevada Test Site.

flow is augmented by contributions from the blocks have generally adopted the "double-porosity" concept proposed by Fissured or fractured rock formations have been the subject Barenblatt et al. [1960]. This concept has been used extenof intensive investigation in recent years. Many productive sively in the petroleum literature. Two approaches to probfreshwater-bearing reservoirs, as well as geothermal and pet- lems of well test analysis have been taken that differ in the roleum reservoirs, are known to be composed of fractured manner by which flow from a block to fissure is described. rock. Also, the search for safe repositories for hazardous One approach assumes that flow occurs under pseudo-steady wastes has led to studies of low-permeability rock formations, state conditions [Warren and Root, 19631 and the other apwhere the major concern is the eventual egress of contami- proach assumes that flow occurs under fully transient connants to the ecosphere through interconnected conductive fis- ditions [for example, Kazemi, 19691. The assumption of sures. pseudo-steady state block-to-fissure flow is an approximation In order to quantify fluid flow behavior in fractured rock that has the advantage over the transient flow assumption by reservoirs, theoretical, laboratory, and field studies have been providing greater mathematical simplicity; however, it has the undertaken. Fractured or fissured reservoirs are complex, het- disadvantage of ignoring some of the physics of the problem. erogeneous, and anisotropic systems. If they are to be treated The assumption of transient block-to-fissure flow is clearly mathematically, certain idealizations are imperative. Under superior from a theoretical standpoint. Interestingly, well test circumstances where the rock matrix has very low per- data exist that suport both approaches. meability, one idealization has been to assume that flow can It is the purpose of this paper to provide a resolution to this occur only in the fractures and not in the matrix. This as- apparent conflict. This is accomplished by using the concept sumption has been used in studies of discrete fractures and has of fracture skin, a thin skin of low-permeability material, deled to the important finding that discharge through a given posited on the surfaces of the blocks, that serves to impede the fracture is proportional to the cube of the fracture aperture. free exchange of fluid between the blocks and fissures. The This "cubic law" appears to hold true for a variety of geom- concept of fracture skin is similar to the idea of fracture etries and for a wide range of stress conditions [witherspoon "damage" presented by Cinco L. and Sarnaniego V . 11977-J for et a!., 1980; Neuzil and Tracy, 1981; Tsang and Witherspoon, wells intersecting vertical fractures. The effect of fracture skin 19813. The assumption of no flow in the matrix is also used in in double-porosity systems is to delay flow contributions from studies of fractured rock masses where intersecting permeable the block to fissure and give rise to pressure responses that are fractures are treated as a "single-porosity" continuum. Be- similar to those predicted under the assumption of pseudocause the fractures often have preferred orientations, aper- steady state flow. By reducing gradients of hydraulic head in tures, and spacings, quantiative studies may require that hy- the compressible blocks, fracture skin provides theoretical jusdraulic conductivity be treated as a second-rank tensor rather tification for the pseudo-steady state flow approximation used than a scalar quantity [Snow, 19691. A general method for in the Warren and Root [I9631 model. To illustrate applifield determination of the hydraulic conductivity tensor of a cation of the model an analysis is made of well test data from fractured rock has been developed by H i h [I9831 and is a pumped well and from an observation well in fractured se called the "cross-holewtest method. volcanic rock at the Nevada Test Site. Studies of fluid flow in a fractured rock mass where fissure
This paper is not subject to U.S.copynght. Published in 1984 by the American Geophysical Union. Paper number 4W0381.

The concept o double porosity was originally proposed by f Barenblatt et ul. [I9691 in order to help quantify flow in fractured rocks. According to this concept, a fractured rock mass

..

Designation of fissure system hydraulic conductivity as shown in (1) has a distinct advantage, as it becomes unnecessary to specify the individual fracture hydraulic conductivity or aperture. In like manner, specific storage of the fissure system can be defined as
Ss = s*, Vf

1
4

(3)

and the specific storage of the blocks can be defined as

Fig. 1. Schematic diagram of a double-porosity reservoir of thickness I I showing a typical representative elementary volume (REV)for the fissure system. is assumed to consist of two interacting, overlapping continua: a continuum of low-permeability, primary porosity blocks and a continuum of high-permeability, secondary porosity fissures. Such a reservoir is depicted schematically in Figure 1. The primary porosity blocks, where the bulk of the fluid is stored, have hydraulic properties that are generally controlled by depositional and lithification processes. The hydraulic properties of the fissure system are generally the result of thermal stresses and tectonic processes. Blocks and fissures may both have been influenced by chemical precipitation and solution processes or by hydrothermal alteration. Because of the dual nature of double-porosity reservoirs and because two controlling partial differential equations may be involved, it is helpful to consider two representative elementary volumes (REV's) in describing the system, one for the fissure system and one for the block system. The fissure system REV is assumed to contain a large number of fissures and blocks (see Figure 1) so that adding or subtracting a few blocks from the REV will not substantially alter its hydraulic properties. The fissure system REV is dependent upon the hydraulic properties and geometry of the fissure system and upon the scale of the probiem under consideration. The block system REV will not necessarily be the same as the fissure system REV and will depend upon characteristics of the blocks. Under the assumption of pseudo-steady state blockto-fissure flow the two REV's can be taken to be equal to one another. Under the assumption of transient block-to-fissure flow the block-system REV will necessarily be significantly smaller in order to account for the distribution of hydraulic head in a representative block. In this paper the block and fissure continua are assumed to be homogeneous and isotropic with regard to their hydraulic properties. Consequently, the REV's will not vary spatially within the reservoir. As described by Gringarten [1982], it is possible to define the fissure system hydraulic conductivity as

where S,, and Ss, are the specific storages of representative fissures and matrix rocks, respectively. The controlling differential equation for flow in the fissure network is assumed to be described by the familiar groundwater diffusion equation with a source term to account for contributions from the matrix rock:

Similarly, for the matrix rock,

dh' K ' V * ~ = S,' - - q, '
St

The primed quantities that appear in this paper refer to the matrix rock and the unprimed quantities refer to the fissure system. Symbols are defined in the notation section. Derivations of the groundwater diffusion equation are given in standard texts such as Freeze and Cherry [1979]. The assumptions under which (5) and (6) are derived include the validity of Darcy's law for flow in the fissures and blocks, slightly compressible fluid and rock, constant hydraulic properties, and negligible effects of fluid density gradients. In their mathematical development for flow to a well in a double-porosity system, Warren and Roor [1963] assumed that the left-hand side of (6) was zero. This means, in effect, that spatial variation of hydraulic head gradients or the divergence of flow in the block is ignored. They also assumed that the flux of fluid from blocks to fissures in an REV occurred in response to the difference in the average hydraulic head in the fissures and the average hydraulic head in the blocks. This is the assumption of pseudo-steady state flow and is described mathematically as

K = KJV,
and the block system hydraulic conductivity as

(1)

K' =: K,Vm

(2)

where K f and K , are the hydraulic conductivitits of representative fissures and matrix rock, ~tspectively, is the ratio Vf of the total volume of the fissures to the bulk volume of the rock mass (the sum of the volume of the fissures and the volume of the matrix), and Vmis the ratio of the total volume of the matrix r o c k to the b u l k , ~ ~ h m ~ and V' sum to 5. unity. Normally, Vm is very c l d t o unity SO that K' 2 K,.

where h' in this case represents a spatial average of hydraulic head within the block and a relates to the geometry of the fissured rock and has the dimensions of inverse area. Barenblatt et al. (19601 made an additional assumption that fluid storage in the fracture network can be ignored, so that the first term on the right-hand side of (5) is negligible compared with the other terms. Because of the low hydraulic conductivity of the matrix rock, the problem of flow to a production well is usually solved by assuming that fluid enters the well bore only through fissures and not through the matrix rock. By assuming radial flow to a fully penetrating well discharging at a constant rate from an inlinitely extensive double-porosity reservoir confined above and below by impermeable formations, the Laplace transform, line source solution for dimensionless drawdown in the fissures can be written as

where 4, =
110

+ pli.

4, is the dimensionless flow from block to fissure in Laplace space, p is the Laplace transform variable, and r , is' the radius of the production well. The well radius is introduced in the mathematical development of (8) for convenience; it is not meant to imply here that the well bore has a finite radius. Provided that block geometry is known, the above doubleporosity system is completely specified by only two dimensionless parameters, (s and i.. Dimensionless drawdown is defined as
I

h, = - - h) (hi Qr

4xK H

Fig. 3. Line source type curve comparisons for transient flow from sphere-shaped blocks (curve A, r,'y2 = 0.01) and slab-shaped blocks (curve B, rD2y2 = 0.01) and pseudo-steady state from shaped blocks (curve C, rD2i. ='0.03). Also shown (crosses)are va ues for sphere-shaped blocks, assuming that ( r D Z y 2 = 0.01:9. )

dimensionless time (inversely related to the Laplace transform variable) is defined as

and dimensionless distance is defined as

, where H is the reservoir thickness, Q is the total well discharge, and h, is the initial hydraulic head in the reservoir. It should be noted that in the line source solution given by (8), t , and rD are not independent of one another and the solution is a function of t D / r D 2 . Although expressed in different notation, (8)-(9) is the same solution as given by Kazemi et a[. C19691, which is an extension for interference tests of the Warren and Root model. If 4, is zero, (8) reduces to the familiar Theis [I9353 solution. Basic to the derivation of (8H9) is the pseudo-steady state flow assumption. Models that are based on this approximation show that, in the absence of well bore storage effects, well discharge consists initially of fluid derived primarily from storage in the fissures followed, at late time, by fluid derived primarily from storage in the blocks. At early and late times, drawdown should therefore follow the familiar Theis type curve. During a sufficiently long transition from early to late time, however, drawdown will approach a plateau (see Figure 3, curve C). Well test data exist that support this assumption [for example, Streltsova, 1976, Figure 7 ; - 3 o w d e tand Gringarten. 19801. Because of such data and because of the relative simplicity of the approach, double-porosity models that

invoke the assumption of pseudo-steady state flow have received much attention in the literature [see Gringarten, 19821. The assumption of pseudo-steady state flow, however, does not have a firm theoretical justification. In order to account for transient flow from blocks to fissures, it is necessary to specify block geometry. Kazemi [I9691 assumed transient block-to-fissure flow and used a finite difference model to simulate flow to a well. He assumed the fractured rock mass could be idealized as alternating layers (slabs) of blocks and fissures where the thickness of the blocks and the aperture of the fissures represent average fracture spacings and apertures. By simulating flow in the blocks as well as in the fissures, he was able to account for the variation that must occur in the hydraulic head in the blocks. This problem was later solved analytically by Boulton and Streltsova [I9771 and Najurieta [I9801 by assuming strictly one-dimensional flow in the block perpendicular to the block-fissure interface. In the notation of this paper, and referring to the schemac diagram in Figure 2 4 the Laplace transform line source solution for dimensionless drawdown in the fissures using slabshaped blocks is (8), where

ql,

= y2m tanh (m)

(15)

A
BIO~;
b'

Fig. 2 Geometrical configuration for (a) slab-shaped blocks and (b) sphere-shaped blocks.

thickness of the 'lab. and b' is the average Type curves for this model differ from those obtained under the pseudo-steady state flow approximation during the period of tiansition from early to latitime. Under the assumption of transient flow there is a gradual transition resulting in early departure from the initial Theis type curve (see Figure 3, curve B). Comparisons of the type curves obtained for each of these two assumptions are given by Streltsooa [1983], Deruyck et al. [1982], and Gringarten [19@]. During a transition period of sufficiently long duration (large a) a semilogarithmic plot of drawdown versus time may show a straight line with a slope of one half the slope that occurs during the late time period (see Figure 6c for S, = 0). Cinco L and Samaniego V . [I9821 pointed out that many data sets published in the literature support the assumption Of transient flow from blocks to sures [for example, Warren bird Root, 1963; Najurieta, 1980

10-1

100

101

lo2
tD/r;

~ 0 3

lo4

105

106

Fig. 4. Type curves showing effects of well bore storage in production well, assuming transient flow from slab-shaped blocks.

class of groundwater and petroleum engineering problems. Moench and Ogata [I9841 provided a discussion of the algorithm and applied it to other problems of groundwater hydrology. In the analysis of production well data and sometimes for the analysis of nearby observation well data it may be necessary to consider effects of well bore storage and well bore skin [Ramey and Agarwal, 1972; Sandel et al., 1978; Chu et al., 19801. The Laplace transform solution for a production well in a fractured reservoir assuming pseudo-steady state flow from blocks to fissures was given by Mavor and Cinco L. [1979]. In the notation of this paper the solution is written as 2[Ko(x) + xS,K l(x)l (19) h,, = P{PWDCKO(X)xS,.Kl(x)l f XKI(X)} + where For an observation well the solution is h, = 2K0(r,s) P(PWDCK~(~)xS,Kl(x)] +

Analytical studies have not been restricted to flow within and adjacent to horizontal slabs. Other geometries, such as cylinders or spheres, have been used to idealize columnar joints, cubes, or parallelepipeds [de Swaan O.,1976; Kuruk 19821. In the notaand Sawyer, 1980; see also Deruyck et tion of this paper, and referring to the schematic diagram in solution for diFigure 2b, the Laplace transform line mensionless drawdown in the fissures using sphere+haped blocks is (8), where
q, = 3yZ[m coth (m) - 11

+ xK,(x))

(20)

S, isthe dimensionless well bore skin commonly used in petroleum engineering. The skin is assumed to be infinitessimally thin, incompressible, and invariant for the duration of the well test. WDis the dimensionless well bore storage, defined in this paper as
C

(18) where C is the well bore storage and is assumed to be constant for the duration of the test. Ramey and Agarwal [I9721 pointed out that the effects of well bore storage can be by virtue of changing liquid level in the well casing or by virtue of compressibility. Effects of well bore storage are greatest when due to changing liquid level. In this case, C = nrc2,where r, is the radius of the well casing in the region of the changing liquid level. If the effects are due to compressibility of the liquid and of the well itself in a pressurized well [see Neuzil, 1982; Bredehoefi and Papadopulos, 19801, C = V,p,gC,,,, where V, is the volume of liquid in the pressurized section, p , is the liquid density, g is the acceleration of gravity, and C,, is the observed compressibility of the combined fluid-well system. The solutions given by (19) and (20) apply equally well when 4, is defined as in (9), (15), or (18). As with the line source solution (8) the above equations can be inverted numerically by using the Stehfest algorithm. If S, and 4, are zero, (19) and (20) reduce to the large-diameter well equation of Papadopulos [I9671 and Papadopulos and Cooper [1967]. If S, and W, are zero, (19) reduces to the solution presented by Warren and Root [1963] for a production well of finite diameter without well bore storage or skin. Figure 4 illustrates typical type curves for a production well, showing the effect of well bore storage by using (19) together with (15) for slab-shaped blocks. Note that with suficiently

and 6' now represents the average radius of the sphere-shaped blocks. Comparisons of the type curves resulting from the different models (slabs, cylinders, and spheres) for transient flow from blocks to fissures show that they are all very nearly the same. Evaluation of the parameters will give different results, however, as can be seen by comparing (15) and (18). For large m (early time), tanh (m) in (15) approaches unity and q, y ( 0 ~ ) ' ' ~From (18) it can be seen that 4, -U 3y (ap)ll'. It is . therefore not surprising that computations show that if the hydraulic conductivity of the slab-shaped block is taken to be 9 times that of the sphere-shaped blocks, the resulting type curves will be indistinguishable at early time. Figure 3 shows a comparison of line-source type curves for sphere-shaped blocks (curve A) and slab-shaped blocks (curve Bf using the indicated values of the parameters a and y. Figure 3 (curve C) shows the corresponding type curve in the case of pseudo-steady state flow from fissure to block where, for slabshaped blocks a = 3/(b1)' [see Warren and Root, 1963, Appendix A]. The effect of reducing hydraulic conductivity in sphere-shaped blocks by a factor of 9 is also illustrated in Figure 3. Although it is not apparent in Figure 3, there is a slight difference in the shape of the type curves during the time of transition to the late time Theis curve for sphere-shaped blocks compared with slab-shaped blocks. This difference becomes more apparent as ( r , ~ )is increased. The curves are ~ bounded on the left by the Theis type c u m and on the right by the Theis type curve displaced to the right by a factor of 1 + a.As (r,y)%r rD2Lis reduced, the type LNNW A, B, and C will move u ~ w a r d Figure 3. Comparison of (9) and (15) or in (18) shows that the equations becorneldentical at'& timk. ' The type curves in Figure 3 were obtained through numerical inversion of (8). using (9). (IS), and (18). This was accomplished by use of the Stehfest [I9701 algorithm, which is easy to use and has proven to be exceptionally accurate for a large

lock b t
Skin

-f7

---------------K
hYy,t,

Fisarre

Fig. 5. Schematic diagram of a block and fissure with fracture skin.

large dimensionless well bore storage the transition from early time to late time is obliterated. RESERVOIRS WITHFRACTURE SKIN FISSURED It is difficult to justify, on theoretical grounds alone, use of the pseudo-steady state block-to-fissure flow model. In response to changes in hydraulic head in a fissure there will be induced in the blocks, near the block-fissure interface, a curvature in the hydraulic head distribution that will increase with decreasing block hydraulic conductivity. This means that during the period of transition from early to late time, when flow from the block has a profound influence on the drawdown in the fissure system, neglect of the left-hand side of (6) is not justifiable. For this reason the transient block-to-fissure flow model which does not neglect the divergence of flow in the blocks has a distinct theoretical advantage. On the other hand, it has been pointed out that well test data exist that support the pseudo-steady state flow model. In this section the transient flow model is modified by introducing the concept of fracture skin. This not only provides theoretical justification for the use of the pseudo-steady state flow model but also allows for evaluation of the fracture skin parameter and, if the data permit, block hydraulic conductivity. It is shown that if the fracture skin parameter is sufficiently large (low fracture skin hydraulic conductivity), most of the change in hydraulic head that occurs in a block occurs across the fracture skin. Consequently, changes in the gradients of hydraulic head in the block become small, thus justifying the assumption in the pseudo-steady state flow model that the divergence of flow in the block is negligible. By suitable modification of the parameter a in (1 I), so that it accounts for the hydraulic conductivity of the fracture skin rather than the block hydraulic conductivity, it is shown that the transient flow model reduces to the pseudo-steady state flow model if the fracture skin parameter is sufficiently large. Figure 5 shows a schematic diagram of a thin but finite thickness skin on the surface of a hypothetical block in a double-porosity reservoir. It is assumed that the skin posseses negligible storage capacity. Flux of fluid from the block to the fissure is assumed to be perpendicular to the interface and to obey Darcy's law. Hence, by continuity, at the fissure-block interface,

Fig. 6b. Comparison of Figure 6a with corresponding type curves for the pseudo-steady state flow model.

where (h'),,,, is the hydraulic head in the block at the skin

surface, K , is the hydraulic conductivity of the skin, and b, is the thickness of the skin. In what follows it is assumed that b, << b'. It is not necessary that K, < K,. Equation (22) derives from heat flow theory [see Carslaw and Jaeger, 1959, p. 201. Fracture skin may occur in naturally fractured reservoirs as a result of mineral alteration or deposition. Such alteration products and minerals are commonly seen lining or filling fractures in surface outcrops or well core samples. Their presence is probably due to the fact that interconnected fissures serve as primary conduits for mineral-charged water circulating naturally in groundwater and geothermal reservoirs. Such fluids may not be in chemical or thermodynamic equilibrium with the wall rock and hence may alter or deposit minerals there. The particular products of hydrothermal alteration or deposition depend upon physical and chemical parameters that vary from place to place. Specific types of such products that occur in geothermal systems were described by Ellis and Mahon [1977]. In a study of the Salton Sea geothermal field, Tewley [1977, p. 321 observed open fractures that ". . . manifest themselves as planar surfaces coated with a thin layer of fracture-filling mineral." Tewley cited evidence that filled or sealed fractures can be reopened by tectonic stresses and natural hydraulic fracturing. Many fractures occurring in the volcanic and carbonate rocks of the Nevada Test Site, Nevada are reported to be filled or partially filled with various mineral assemblages [Young, 1972; Blankennagel and Weir, 1973; Winograd and Thordarson, 1975; Spengler et al., 19791. Excellent photographs of cores with partially filled fractures are presented by

Fig. Q. Line source, type cuwes'for transient f o from slab-shaped lw blocks with fracture skin.

Fig. 6.2. Semilogarithmic plot of Figure Q.

For sphere-shaped blocks,

v/ 2 v,
hence

4 ~ ( b ' ) ~ b 36 4/3n(b' + b)3 4/37r(b')3 4/3n(br+ b)'

b '

(29)

(30)

(31)

0.2

0.4

0.6
D

0.8

The initial condition is

h=hi
r,sr

Fig. 7a. Distribution of hydraulic head for various values of S, in slab-shaped blocks at the indicated values of dimensionless time and distance.

~h~ boundary

are

Ducidson and Snowdon [I9781 for a carbonate rock gas reservoir in northwestern Canada. It is reasonable to conjecture that such mineral deposits are common in nature and that many serve to plug pores of primary porosity blocks at fissure block interfaces. Transient Block-to-Fissure Flow In this section the boundary value problems and Laplace transform solutions are given for transient block-to-fissure flow with fracture skin. Derivations for slab-shaped blocks and sphere-shaped blocks are given in Appendices A and B; see Figures 2 and 5 and the notation section for definition of symbols. The generalized coordinate y in Figure 5 becomes z for slab-shaped blocks and p for sphere-shaped blocks in what follows. The controlling differential equation for flow in a fissure is assumed to be the radial diffusion equation with a source term,

dh, 4nr.~~bne) =nQf+CX

,=,Y

r=rw

(34)

where

Q r is the constant discharge from a well intersecting a single fissure of aperture 2b, and n is the number of producing fissures. Replacing n by H/2b' in (34) and assuming that b' + b b', the well bore boundary condition becomes

where

a2h 1 dh Kr-+Kr--;-=S ar

dh

at + 9

r, I r

(230)

Slab-shaped blocks. Referring to Figure 2a, the governing equation for slab-shaped blocks is the planar diffusion equation

where
(23b) By substituting (1) and ( 2 ) into these expressions, the controlling equation for the fissure network becomes a2h dh qVr - + -l - = - -S , a h +dr2

a2h' -.=-c?z2
The initial condition is

S ah' " K' at

O < z ~ b '

(38)

h'=h,

O < z ~ b '

(39)

r ar

K at

rw 5 r

(24)

where
(25)

For slab-shaped blocks,

b b '-bl+b-b'

----

(26)

v = -Y 1 , b 'b'+ b
hence

(27)

)K " ( = i ! z

Kb'

a2 = ,,

(28)

Fig. 76. Distribution of hydraulic head for various values of SF in sphere-shaped blocks at the indicated values of dimensionless time and distance.

) R

FISSURED GROUNDWATER RESERVOIR

TABLE 1 . Well Characteristics
Pumped Well Well Number Ground elevation, m Total depth, m Approximate depth to static water level, rn Well diameter below static water level. m Drilling medium Bottom of casing, m Perforations, rn UE-25b# 1 1198.7 1219.2 470 0.22 air. detergent, and water 518 477-501

837

*heboundary conditions are

Observation Well UE-25a# 1 1198.7 762.2 470 0.075 bentonite mud none none

vhere
S, = K'h, jK,h' (42)

:quation (41) is a rearrangement of (22) with K' 2. K,. SF is ermed the dimensionless fracture skin in this paper. In addition to the dimensionless parameters in ( 1 2 H 1 4 ) the allowing are defined: h,,
= - - hw,) (hi

4nKH

QT

hD ' - - 4 n K H (hi - h ' ) QT

Sphere-shaped blocks. Referring to Figure 2b, the governing equation for sphere-shaped blocks is [see Carslaw and Jueger. 1959, p. 2301 ?2(ph') ----- S,' PP2
c'(pht )
(71

The above coupled boundary value problem for slab-shaped blocks is put in dimensionless form by incorporating the dimensionless parameters and is solved in Laplace space (see Appendix A). The Laplace transform solutions for dimensionless hydraulic head in the production well and in the fissure network is given by (19) and (20) with

K'

0c:psb'

The initial condition is

hf=hi O<pIbl (49)

The boundary conditions are h' finite

p =0 (50)

q0 = 1 + S,m tanh (m)

?'m tanh ( m )

(46)

hf = h - blsp(ahf/ap)

p = b1

(51)

The Laplace transform solution for dimensionless drawdown within a representative slab-shaped block is cosh ( z g ) [l + S,m tanh (m)] cosh (m)

With one additional dimensionless parameter defined as

AD

P, i
(47)

= pi'b'

(52)

For a line source the Laplace transform solution is given by (8) with qD as defined in (46). The line source solution is not derived in this paper but can be obtained in a manner similar to that outlined in Appendix A by having r - + O and by making appropriate changes in the inner boundary condition
(36).

the above coupled boundary-value problem is put in dimensionless form and solved in Laplace space (see Appendix B). The Laplace transform solutions (19) and (20) also apply for

2 0

1200

40

60

80

PERCENT OF TOTAL FLOW

Fig. 8. Typical mineral-filled fractures. (Top) Core from well UE-25a# 1 at a depth of 676 m. (Bottom) C o n from well UE-25bA 1 at adepth of612 m.

Fig. 9. Bore hole survey showing percent of total flow versus depth and stratigraphic units of Miocene age in pumped well UE-25b # 1 (adapted from Lahoud et al).

TABLE 2 Drawdown Data for Test 3 .

t,

TABLE 2 (Continuedl .
t , min

min
Pumped Well

hi - h, m
2.513 3.769 4.583 4.858 5.003 5.119 5.230 5.390 5.542 5.690 5.960 6.19 6.42 6.59 6.74 6.96 7.17 7.33 74 .5 7.56 7.76 7.93 8.03 8.12 8.24 8.32 8.41 8.46 8.54 8.62 8.67 8.70 8.74 8.76 8.77 8.81 8.84 8.84 8.86 8.86 8.90 89 .1 8.92 8.93 8.95 8.97 8.98 8.99 9.00 9.02 9.04 9.07 9 11 . 9.14 9.17 9.18 92 .1 9.25 9.30 9.44 9.55 9.64 9.74 9.78 980 9.84 99 .3 10.03
10.08

h,

- h, m
10.30 10.4 1

Pumped Well (Continued)

4000.0 4200.0
Obseruation Wl el

1.6 02

0.5 0.6 0.7 0.8 0.9 1. O 12 . 14 . 1.6 18 . 2.0 2.2 2.4 26 . 2.8 3.0 3.5 4.0 4.5 5.0 5.5 6.0 7.0 80 . 9.0 10.0 1. 20 14.0 16.0 18.0 20.0 25.0 30.0 35.0 40.0 45.0 50.0 60.0 70.0 120.0 140.0 160.0 180.0 200.0 240.0 280.0 300.0 340.0 400.0 440.0 500.0 600.0 700.0 800.0 900.0 1000.0 1200.0 1400.0 1600.0 1800.0 2000.0 2300.0 2700.0 3000.0 3500.0 3680.0

0.002 0.002 0.002 0.002 0.002 0.005 0.005 0.007 0.007 0.012 0.015 0.015 0.020 0.022 0.025 0.027 0.037 0.045 0.052 0.059 0.069 0.079 0.097 0.116 0.134 0.151 0.186 0.213 0.238 0.260 0.285 0.320 0.342 0.359 0.374 0.384 0.392 0.401 0.411 0.434 0.439 0.444 0.45 1 0.453 0.461 0.468 0.471 0.478 0.491 0.498 0.506 0.518 0.525 0.528 0.528 0.538 0.563 0.577 0.577 0.577 0.590 0.587 0.615 0.615 0.627 0.639

there-shaped blocks, with

TABLE 3. Match Point and Dimensionless Parameters

40 = { I

3y2[m w t h (m) - 11 SF[m coth (m) - 11)

(53)

Parameter
h, - h WD
..z
tDlrD2

Value
1.4 1.0 m 1 lo3 1.8 x

h~

he Laplace transform solution for dimensionless drawdown any point within a representative sphere-shaped block is
h~ sinh (p,m) = p D { l + SF[m coth (m) - I]) sinh (m) (54)

t/r2
0

As for slab-shaped blocks the line source Laplace transform ~lution given by ( 8 ) with qDnow defined by (53). is
seudo-Steady Stare Block-to-Fissure Flow In this section the approach of Warren and Root [I9631 and f other investigators that invokes the pseudo-steady state ow assumption is modified to incorporate fracture skin. Jhen this is done the hydraulic conductivity of the block is :placed by the hydraulic conductivity of the fracture skin, :suiting in a revised definition of i Evaluation of 2. from a . . jell test will then yield a value of the ratio K,/b, rather than ydraulic conductivity K'. Derivations of the Laplace trans)rrn solutions are given in Appendix C. Substituting (22) and (25) into (24) and comparing the result ith (5),it is seen that (7) can be written as

s,

4 x lo8 106 min rn-2 2 x lo2 1.o

and for sphere-shaped blocks

(58) and (59) can be used in (8), (19), or (20) to obtain the necessary type curves. Comparison of the Laplace transform expressions (46) and (53) for the transient flow assumption with the expressions (58) and (59) for the pseudo-steady state flow assumption shows that they become equivalent for certain ranges of the indicated parameters. For large SF and large m (short time), (46) and (58) become

q,=

b bs .here h' - h is the head differential across the skin of thickess b, and V, has been assumed equal to unity. In this develpment it is not necessary to specify the location of h' in (22) s ( l ~ ' ) because it is assumed that h' is uniform throughout ~ ~ ~ , le block at an any instant of time. Accordingly, it is now dmissible to neglect the divergence of flow in the blocks, and i) becomes
q = S f Iat

-v,-- K , (h' - h)

and (53) and (59) become

For large SF and small m (long time), (46) and (53) become identical to (58) and (59).This occurs when m2 5 0 1or, in real . space, when the following criterion holds:

ah'

Alternatively, the criterion can be written as

Combining (5), (55), and (56), it is shown in Appendix C that is given by ( 9 ) with

I = -rw 2K."f
Kbb,

y2 V f b f SF b

hus, for slab-shaped blocks,

Hence, for the criterion (62) to hold over a broad time range, block diffusivity ( K f / S S ' ) must be large and/or block dimensions must be small. For a given block size, (46) and (53) will reduce to (58) and (59) only if SF is large and block diffusivity is large or, alternatively, if the time of interest is large. When SF is zero, (46) and (53) reduce to (15) and (18). For the pseudo-steady state flow assumption, however, zero SF implies immediate and complete drainage from the blocks due to any decline in hydraulic head in the fissures. In this instance A expressions for qD approach that given in (64) when m is U sufficiently small. Figure 60 shows line source type curves using (8) and (46) for different values of SF and the indicated values of a and (rDy)'. Figure 6 b shows the same type curves compared with the corresponding curves using (9) or (58), where 1 is obtained by dividing y2 by S,. As SF increases, it can be seen that the type curves obtained by using (44) approach those using (58). A similar correspondena can be shown using sphere-shaped blocks. As was pointed out in the last section, where SF = 0, a

ig. 10. Drawdown data for test 3 in well UE-2%# 1 (r, = 0.11 m) and well UE-25a X 1 (r = 110 m . )

MOMCH: DOUBLE-POROSITY FU MODELS

TABLE 4. Calculated Hydraulic Properties

Parameter

Value

semilogarithmic plot of drawdown versus time during the transition from early to late time may show a straight line with a slope equal to one half that of the late time semilogarithmic straight line. This is shown in Figure 6c, which is the same as Figure 6a plotted in semilogarithmic coordinates. In the presence of fracture skin there may appear to be a semilogarithmic straight line, but its slope may be very much less than one half the late time slope and. if not properly interpreted, may gibe rlse to an erroneously large computed value of fissure system hydraulic conductivity. Figures 7a and 7b show the distribution of hydraulic head, computed using (47) and (54), for different values of S , in slab-shaped and sphere-shaped blocks at a given time and given radial distance from the production well. It is apparent in both instances that the distribution of hydraulic head in the block becomes uniform when SF is sufficiently large. This then provides the necessary theoretical justification for neglecting the divergence of flow in (6) and for use of the assumption of pseudo-steady state flow from block to fissure.

In order to assist the U.S. Department of Energy in determining the suitability of underground respositories for storage of waste radionuclides, the U.S. Geological Survey has been conducting intensive investigations t o characterize the hydraulic properties of the Tertiary volcanic rocks in the vicinity of Yucca Mountain at the Nevada Test Site. Geologic, geophysical, petrologic, and hydrologic studies have been made, and numbers of wells have been drilled, cored, and tested. Scott et al. [I9833 gave details of the geologic setting, structure, stratigraphy, rock physical properties, and fault and fracture systems in the area of interest known as Drill Hole Wash, located on the east flank of Yucca Mountain. For the present study, well test data were made available (J. Robison, personal communication, 1982) from a pumped well WE-25b# 1) and a n observation well (UE-25a# 1). These wells were drilled through thick sequences of fractures and faulted nonwelded to densely welded rhyolitic, ash flow, and bedded tuffs to depths below the water table which is located at a depth of 470 m. Circulation losses that occurred during wring of the highly fractured intervals suggest that many fractures are open and interconnected [Spengler et al., 19791. Core samples reveal that most of the fractures are steeply dipping and coated with deposits of silica, manganese and iron oxides, and calcite. Down-hole televiewer records show that water-producing zones are associated with low-angle fractures (R. G. Lahoud et al., unpublished manuscript, 1984; hereinafter Lahoud et al.). Figure 8 shows photographs of typical mineral-filled fractures in cores taken from the observation wdl and pumped well at depths of 676 m and 612 m, respectively. Rock at these depths was reported t o be partially t o mdderately welded, devitrified ash flow tuff belonging t o the Prow Pass Member of Crater Flat Tuff of Miocene age [Spengler et al, 19791. The minerals deposited in these fractures probably indude MnO, and silica.

Further details regarding the fractures, rocks, and minerals encountered in these boreholes were provided by Spengler er a/. El9791 and Caporuscio et al. [1982]. Lahoud et al. provided specific information about the pumped well including construction, geophysical logs, borehole flow surveys, results of packer and pumping tests, water chemistry, fracture analysis, and results of laboratory tests performed on rock cores. Although both wells deviate from the true vertical, they appear to deviate by about the same magnitude and direction. The contacts between the producing formations occur at about the same depths in the pumped well and in the observation well. For these reasons, no corrections were applied in calculating the distance between zones of entry in the two wells. Pertinent information regarding the wells is given in Table 1. The results of a bore hole Row survey conducted in the pumped well are shown in Figure 9. There appear to be five major zones of entry over a depth interval of about 400 m, which is taken to be the reservoir thickness for the purpose of this analysis. The lack of significant head differences in packed-off intervals under static conditions suggests good hydraulic connection between zones (Lahoud et al.). Table 2 shows the drawdown data for test 3 that are analyzed in this case study. Figure 10 shows a logarithmic plot of h versus t / r 2 where r , = 0.11 m for the pumped well and r = 110 m for the observation well. Test 3 was the third test in a series of three that were conducted on well UE-25b# 1 in October 1981. Test 1 was run as a single-well test with a small capacity pump. The borehole flow survey was conducted at the end of this test. Test 2 was a long test (9 days), but the flow rate was not constant and the drawdown curves show the effect of continuing well development (D. Lobmeyer, personal communication, 1982). During the final two days of test 2 the well was pumped at the same rate as test 3. It is encouraging to note that recovery data from test 2 are nearly identical to drawdown data for test 3, except for times less than 6 minutes, for both the pumping well and the observation well. Test 3 was pumped a t a constant rate of 35.8 L/s for nearly 3 days. At the end of this test the pumped water had a temperature of 35C at the well head. Plots of drawdown and recovery for these various tests were given by Lahoud et al. It was decided to analyze the drawdown data by assuming slab-shaped blocks. This was due to the scale of the problem and the observation that the distance between the two wells and the average distance between the zones of water entry shown in Figure 9 are of the same magnitude, about 100 m. In

,
I
,

Fig. 11. Comparison of drawdown data to line source type curves using the pseudo-steady state block-to-fissure flow model and the indicated parameters.

the absence of detailed description of the fracture network, the believes that more closely spaced water entries are to justify the use of sphere-shaped blocks. Six dimensionless parameters (see Table 3) can be obtained by a type curve match of the data in Figure 10. From these, reservoir parameters can be estimated. Fortunately, these dimensionless parameters tend to influence the character of different parts of the type curves. This makes it possible to obtain a set of parameters that is unique to the problem at hand. Nevertheless, it is helpful to use a systematic approach in attempting to obtain a match. In the analysis of the data in Figure 10 the parameters u and 2 were first estimated by . applying the pseudo-steady state block-to-fissure flow model. This involved numerical inversions of the Laplace transform solution for a line source given by (8) and (9) using different values of a and 2.. A rough type curve match was obtained by varying u to obtain a match in the horizontal direction and varying i. to obtain a match in the vertical direction. As shown in Fjgure 11, a single value of o and j gives a rough match to . both the pumped well data and the observation well data. To avoid confusion, the data in Figure 11 (and in Figures 12 and 13) are plotted without showing the ordinate and abscissa. The data are plotted at the same scale shown in Figure 10, with the match point indicated on the figures. The type curve match shown in Figure 11 under the assumption of pseudo-steady state block-to-fissure flow might be considered satisfactory. Such a match is not possible under the assumption of transient block-to-fissure flow without the influences of fracture skin. Fortunately, because the dimensionless fracture skin factor is not so large as to obliterate the transient character of the flow from the blocks, it is possible to use the theory presented in this paper to obtain separate values of y2 and S,. This will allow for evaluation of K'. This is done as illustrated in Figure 12 by using (8) and (46) and holding the ratio y 2 / S F constant (equal to 2.) while varying S Of the three type curves for r, = 1000 shown in Figure 12, , . SF = 1 appears to provide the k t agreement with the observation-well data. At the resolution of this plot, pumped well data is not diagnostic of S,. When an expanded scale is used, the pumped well data can be shown to match type curves for SF = 1 or greater. The final match between data and type curves is shown in Figure 13. Laplace transform solutions that include effects of well bore storage, given by (19) and (20), were used together with the expression for q, given by (46) for slab-shaped blocks. Well bore skin was assumed to be negligible. This assumption

1 ' Data

Fig. 13. Final type curve match using the transient block-to-fissure flow model with fracture skin and well bore storage. appears justifiable in view of the following facts: (1) the major producing zone in the pumped well contains no casing, (2) the well was drilled using air and detergent as a circulation medium rather than drilling mud, and (3) the well was thoroughly developed by prior production testing. Table 3 lists the values of the match point and dimensionless parameters obtained by the type curve match in Figure 13. The reservoir hydraulic properties are calculated from these values and are listed in Table 4. In calculating the fissure system hydraulic conductivity, the reservoir thickness, estimated from the bore hole flow survey, was taken to be 400 m. Also, as there were five major zones of water entry, the block thickness 2b' was estimated to be 80 m. The fissure system hydraulic conductivity shown in Table 4 is consistent with values obtained by packer injection tests in the pumped well (Lahoud et al.) and with values obtained for fractured rock reservoirs in other locales [see Freeze and Cherry, 1979; Winograd and Thordarson, 19753.Of the calculated parameters shown in Table 4, K and S, are probably the most reliable, given that the reservor thickness was accurately estimated. Hydraulic conductivity obtained for the blocks is subject to uncertainty, largely because of the ratio r,/bl, which is squared in the calculation. However, it is not inconsistent with values obtained by packer injection tests, given by Lahoud et al., that range from 3 x lo-' to 6 x m/s in intervals containing no major producing fractures. Laboratory measurements of the hydraulic conductivity of cores taken from the wells suggest that matrix hydraulic conductivity is two to five orders of magnitude less than that shown in Table 4 (Lahoud et al.). The value of block hydraulic conductivity obtained by this analysis, therefore, most likely represents the hydraulic conductivity of a system of joint surfaces within the matrix blocks. This suggestion is also supported by noting that the block specific storage shown in Table 4 appears to be two or three orders of magnitude larger than values calculated for unfractured rock [see van der Kamp and Gale, 1983, Table I]. It can probably only be accounted for by the presence of highly compressible microfissures within the blocks. It should be pointed out that use of sphere-shaped blocks rather than slab-shaped blocks would reduce this discrepancy in block hydraulic conductivity by one order of magnitude but would not necessarily mean that sphere-shaped blocks are a more realistic descriptive model. Because it is not possible to determine the average fracture skin thickness, only the ratio K,/b, is given in Table 4. The value of dimensionless well bore storage shown in Table 3 that was obtained by the type curve match in Figure

Fig. 1 2 Comparison of drawdown data to line source type curves asing the transient block-to-fissure flow model with fracture skin and

the indicated parameters.

842

MOENCH: I)OUBLE-POROSITY MODELS FISSURED FOR GROUNDWATER RESERVOIR Figure 9 appears to have been fully penetrated by the pumped well. Also, there is evidence that there is good hydraulic connection between producing zones. It is probable, however, that effects of anisotropy are significant. This is evident from field and bore hole observation of preferred fracture orientation and from the natural tendency of bore holes in the area to deviate from the vertical toward the west and southwest [Scott et a[., 19831. A well test with data from a pumped well and a single observation well is insuflicient to evaluate the hydraulic conductivity tensor. It is also possible that hydraulic boundaries due to major faults or intrusive dikes and sills are present within the flow regime. The change in slope that occurs at t = 1000 min in Figure 14b might possibly be interpreted by taking these factors into account; however, the change in slope is on the order of 10 :I rather than 2 : I , which is characteristic of a single hydraulic boundary. Also, as the data appear to be consistent with the assumptions of the proposed double-porosity model, it is not necessary to call upon added complications. The prevailing theories of flow to a well in a doubleporosity reservoir can be unified by accounting for fracture skin in the mathematical development. When modified to include fracture skin, the commonly used theory based on the assumption of pseudo-steady state block-to-fissure flow is found to be a special case of the theory presented in this paper that is based on the assumption of transient block-to-fissure flow with fracture skin. The former applies when fracture skin hydraulic conductivity is sufficiently low. In the absence of fracture skin the pseudo-steady state flow model does not have a sound theoretical basis. The interference test data from the Nevada Test Site support the hypothesis that fracture skin may be important in some double-porosity systems. Use of the modified theory for analysis of these data shows the advantage of applying the transient block-to-fissure flow assumption: it is possible to evaluate "block" hydraulic conductivity if the fracture skin hydraulic conductivity is not so low as to obscure the transient character of the flow in the blocks. Comparison of the result obtained by this analysis with laboratory measurements of block hydraulic conductivity suggests that the well test data analysis gives rise to an apparent block hydraulic conductivity that includes effects of microfractures within the matrix rock. Caution is urged in the use of the standard semilogarithmic straight line method for evaluating the product of hydraulic conductivity and reservoir thickness in double-porosity systems. Analysis of the data in the report shows that the hydraulic conductivity of the fissure system may be overestimated by one order of magnitude if the straight line method is applied to the observation well data. Because the presence or absence of fracture skin can have marked influence on the exchange of fluids between blocks and fissures, careful consideration of whether or not it is present should be given. This is especially true in regions under study for the disposal of toxic wastes. The presence of fracture skin due to mineral deposition or alteration will not only affect the adsorptive characteristics of the rocks but will also alter the flow velocities in the fractures and, consequently, the dispersive characteristics of the fractured rock system. In this appendix the derivation of the Laplace transform solutions is given for the case of transient flow from slab-

t (mini

Fig.

14u.

Evaluation of KH from a semilogarithmic plot of drawdown data for the observation well.

13 agrees closely with the calculated value of W,. This provides an internal check on the calculated value of S, or the estimated value of reservoir thickness H. Because effects of well bore storage are due to changing liquid level, and as in this test r, = r,, (21) reduces to

The calculated value of W, obtained from (65) is 833. It was pointed out in the previous section that use of the semilogarithmic straight line method for evaluating KH in fractured rock reservoirs should be done with caution. As shown in Figure 14a, application of this approach to the observation well data in this field test produces a nice straight line and results in a value of fissure system hydraulic conductivity that is about an order of magnitude greater than that shown in Table 4. This is consistent with the pnsent analysis if these data do indeed occur during the period of transition from early to late time under the influence o fracture skin. As f shown in Figure 146, a semilogarithmic plot of late time (greater than 1000 minutes), pumped well data also produces a straight line and results in a fissure system hydraulic wnductivity that agrees (within 20%) with that shown in Table 4, as it should. The results of this analysis support the contention that the flow system can be treated successfully with a double-porosity model. By incorporating fracture skin in the model, it appears possible to obtain an improved description of the flow system. Not included in the analysis are effects of partial penetration and anisotropy. The former is probably not important in this well test because the major zone of production shown in

Fig. 14b. Evaluation of KH from a ~ g a r i t h m i c plot of drawCI down data for the pumpal W I .

The subsidiary differential equation for the block system is shaped blocks with fracture skin. The coupled boundary value given by (24), (28), (32), (33), ( 3 3 , and (36) for the a2hD1/azD2 m2hDf o 5 Z , I1 = (~16) fissure system and by ( 3 8 x 4 1 ) for the block system is put in nondimensional form by substituting the definitions for di- The subsidiary boundary conditions are mensionle~~ drawdown, time and distance given by (12)-(14), ahD1lazD o = zD = o ('417) (21), and ( 4 3 x 4 5 ) and the dimensionless parameters a, y, SF, and m. l-b 5hD1 z = 1 , (A181 D - ,-SFFor the fissure system the governing differential equation 82, becomes The general solution to (A11) is a2hD 1 ah ah -+-2 =--A+ 4' , r, 2 1 (Al) h,' = A cosh (z,m) + B sinh (z,m) (A191 drD2 rD ar, at, where
~D'=Y

ah,'
azo

( )

Applying the boundary conditions (A 17) and (A18) the particular solution is h ~ = [I '

,,=I

The initial condition is hD=O The boundary conditions are h,=O and r,=m ('44) where
r D 21

+ S,m

b~ cosh (zDm) tanh (m)] cosh (m)

(A20)
=

(A31

Substitution of (A20) into (A1 I ) and (A12), letting q,'

qDh, yields

7% tanh (m) where

'D

=1

+ SFm tanh (m)

(A20)and (A22)are the solutions (47)and (46). The general solution to (A21)is For the block system the governing differential equation becomes d2hD' = -

where
x = (p

azD2

u ah,' y2 at,

OI;z,sI

(A71

+ qD)ll2
('424)

Since by (A13),h, is bounded, C = 0, and hD = DKo(rDx) OIz,< 1 Applying (A14)

The initial condition is hDf=O The boundary conditions are ahD1lazD 0 = ah,' h,' = h, - SF ~ Z D
2,

=0

z,= 1

By applying ( A l S )and by algebraic manipulation, the particular solutions are

By performing the Laplace transformation procedure on ( A l H A l O ) ,the coupled subsidiary boundary value problem is obtained. The subsidiary differential equation for the fissure system is
(A26)and (A27)are the solutions (19)and (20),where where
x = (p

&)"2

and 4, is defined by (A22). The subsidiary boundary conditions are hD=O

PW~;,,

r,=m / ~ r, = 1

(A131 ( A14)

- (ahD/arD) 2 =

where

In this appendix the derivation of the Laplace transform solutions is given for the case of transient flow from sphereshaped blocks with fracture skin. The coupled boundary value problems given by (24). (31)-(33), ( 3 3 , and (36) for the fissure system and by ( 4 8 x 5 1 ) for the block system are put in nondimensional form, as in Appendix A. Dimensionless distance in the block is defined by (52).

T h e boundary value problem for the fissure system is the same as (A 1HA6) with the exception that (A2) is replaced by

with (S), the controlling differential equation becomes, in radial coordinates,

-+

a2h ar2

--=--

1 ah r ar

S, ah S,' ah' +-K at K at

For the block system the governing differential equation becomes

An additional equation is obtained by equating (55) and (56) so that

,

--=-

S,' ah' K at

KsV' (h, - h,,) ~bb,

The initial condition is

Ir,'=O

01p,1

033)

After substituting the various dimensionless parameters, (C1) and (C2) become
a2hD 1 ahD - - - = A +Sh + arD r, ar, St, odh,' dt,

The boundary conditions are

h,' h,'

(C3)
((241

finite
ih,'

p, = 0
c p ~

(B4)
c
=

-= j.(h,) - hIIr 1

= 1 1 , - SF - ;

p,

(B5) where

Sh,' r't,

The subsidiary boundary value problem for the fissure system is the same as (All)iA15) with the exception that (A12) is replaced by

"2
I

/,'

. '

s, VFh

(C5)

-d2(pd~')
d ~ D 2

(B6)

Performing the Laplace transformation procedure on (C3) and (C4) and combining the results the subsidiary differential equation becomes
-+--=

The subsidiary differential equation for the block system is

- m2(pdD')

a2iD

a);,
P&,

O I PD I 1

(B7) where

arD2 rD arD

+~ D L D

(C6)

The subsidiary boundary conditions are

h,'

finite

pD E 0
P D = ~

(B8) (B9)

4,

up). ap + 2.

(c7)

dii,' hDf= h, - SF d~~

For the slab-shaped blocks, (26) applies, and (C7) becomes y2m2
QD

The general solution to (B7) is

1 + SFm2

(C8)

~ 9 = ' cash (pfl) + ,A

For (B8) to hold, A = 0 and

sinh @fl)

(B1O) For sphere-shaped blocks, (29) applies, and (C7) becomes

B h,' = - sinh (pfl)

PD

NOTATION

.1J

By applying boundary condition (B9), the particular solution sinh @g)

hD' =

p,{l

+ SF[mcoth (m) -.

hD

11) sinh (m)

(B12)

2b 2b' b, C

The subsidiary differential equation for the fissure system is (%I). Again, as in Appendix A, by letting &' = qdD,

C, ,,
g

(Y2)

and (B13) are the solutions (53) and (54). The Laplaa tr nsform solutions for the dimensionless drawdown in the fissure system are given by (A26) & (A27) in Appendix A with qD as defined in (B 13).

H h h' h, h, I . KO
K,

In this appendix derivations are given of the Laplace transform solutions for the case of p s e u d ~ t e a d ystate flow from . slab-shaped and sphere-shaped blocks with fracture skin. By neglecting the left-hand side of (6) and combining the result -

K K'

K,
K,

average aperture of fissures, L. average thickness or diameter of blocks. L. average thickness of fracture skin, L. well bore storage coefficient, L2. observed compressibility of fluid well system, L T ~ M'. acceleration of gravity, LT -2. reservoir thickness, L. hydraulic head in fissure, L. hydraulic head in block, L. initial hydraulic head, L. hydraulic head in pumped well, L. modified Bessel function of the first kind and order zero. modified Bessel function of the second kind and order zero. modified Bessel function of the second kind and order unity. hydraulic conductivity of the fissure system, L T - I . hydraulic conductivity of the block system, L T - l . hydraulic conductivity of an average fissure, L T - '. hydraulic conductivity of an average block, L T - '.

n number of producing fissures. p Laplace transform variable. ischarge t o the pumped well from a n average fissure, Qf d3T-l . L Q , total well discharge, L ~ T - ' . q source term for transient block-to-fissure flow, T - '. q, source term for pseudo-steady state block-to-fissure flow, T - I . r radial distance measured from center of pumped well, L. r , effective radius of pumped well, L. r, internal radius of pumped well casing, L. S, specific storage of the fissure system, LS,' specific storage of the block system, L-'. S,, specific storage of a n average fissure, L-'. S specific storage of a n average block, L-'. , , S , dimensionless well bore skin. S, dimensionless fracture skin. -r t time since start of pumping, T . V ratio of total volume of fissures t o the bulk volume. , V ratio o f t o t a l volume of blocks t o the bulk volume. , V_, volume of liauid in a d ~ r e s s u r i z e well. L3. d W, dimensionless well bore storage. y generalized block coordinate, L. z distance measured from center of a slab-shaped block t o the fissure, L. a relates t o geometry of blocks under assumption of pseudo-steady state flow, L-2. p, density of fluid in the pumped well, ML-3. p distance measured from center of a sphere-shaped block t o the fissure, L. D (as a subscript) dimensionless parameter.
-

K, hydraulic conductivity of fracture skin, LT-

'.

'.

Dimensionless Groupings for Reservoir Parameters

= (r,/b'KK'/K)'12 m = (ap)I1'/y

u = Ss1/Ss

SF = K'b,/K,bf
Acknowledgments. The well test data used in this analysis were provided by personnel of the Nuclear Hydrology Program, U.S. Geological Survey, who are gratefully acknowledged. Photographs of rock core with mineral-filled fractures were taken by John Downing of Pan American World Airways, Mercury, Nevada. Insightful discussions with my colleague Paul Hsieh during preparation of the manuscript led to significant improvement in the content of the paper. Thanks are also due to Garth van der Kamp for many helpful comments and suggestions.

REFERENCES
Barenblatt, G. I., Iu. P. Zheltov, and I. N. Kocina, Basic concepts in the theory of seepage of homogeneous liquids in fissured rocks (strata), J. Appl. Math. Mech. Engl. Transl., 24, 1286-1303, 1960. Blankennagel, R. K., and J. E. Weir, Jr, Gtohydrology of the eastern part of Pahute Mesa, Nevada Test Site, Nye County, Nevada; U S . Geol. S w v . Prof: Pap. 712-B,35 pp, 1973. Boulton, N. S., and T. D. Streltsova, Unsteady flow to a pumped well in a fissured water-bearing formation, J. Hydrol, 35,257-270, 1977. Bourdet, D., and A. C. Gringarten, Determination of fissure volume and block size in fractured reservoirs by typecurve analysis, UnPublished preprint, Pap. SPE 9293, Soc. of Pet. Eng. of AIME, Dallas, Tex, 1980. Bredthoeft, J. D., and S. S. Papadopulos, A method for determining the hydraulic properties of tight formations, Woter Resola. Res., - lql), 233-238,1980. w n r s d o , F D Vaniman, D. Bish, D. Broxton, B. Arncy, G. , . Heiken, F. Byers, R. Gooley, and E. Sumrge, Petrologic studies of drill cores USW-G2 and UE25b-1, Yucca Mountain, Nevada, Rep.

LA-9255-MS, 111 pp., Los Alamos Natl. Lab., Los Alamos, N. M., 1982. Carslaw, H. S., and J. C. Jaeger, Conduction of Heat in Solids. 510 pp., 2nd ed., Oxford University Press, London, 1959. Chu, W. G., J. Garcia-Rivera, and R. Raghaven, Analysis of interference test data influenced by wellbore storage and skin at the flowing well, Trans. Soc. Pet. ~ nAIME,. 269, 171-178, 1980. ~ Cinco L., H., and F. Samaniego V., Erect of wellbore storage and damage on the transient pressure behavior of vertically fractured wells. Pao. SPE 6752. Soc. of Pet. Ene. of AIME. Dallas., Tex... 1977. Cinco L., H., and F. ~arnaneigoV., Fressure transient analysis for naturally fractured reservoirs, Pap. SPE 11026, Soc. of Pet. ~ n g of . AIME, Dallas, Tex., 1982. Davidson, D. A., and D. M. Snowdon, Beaver River Middle Devonian Carbonate: Performance review of a high-relief, fractured gas reservoir with water influx. J P T J. Pet. Technol., 30(12), 1672-1678. 1978. Deruyck, B. G., D. P. Bourdet, G. DaPrat, and H. J. Ramey, Jr.. Interpretation of interference tests in reservoirs with double porosity behavior: Theory and field examples, Pap. SPE 11025, Soc. of Pet. Eng. of AIME, Dallas, Tex.. 1982. de Suaan. O., A., Analytic solution for determining naturally fractured reservoir by well testing, ~ r a n s ;Soc. Pe;. Eng. AIME, 261, 117-122, 1976. Ellis, A. J., and W. A. J. Mahon, Chemistry and Geothermal Systems, 392 no.. Academic. New York. 1977. . Freeze, R. A., and J.' A. cherry.'Groundwater. 604 DD., Prentice-Hall. .. Englewood Cliffs, N. J., 1979 Gringarten, A. C., Flow-test evaluation of fractured reservoirs, in Recent Trends in Hydrogeology, Spec. Pap. Geol. Soc. Am. 189, 237-263. 1982. Hsieh, P. A,, Theoretical and field studies of fluid flow in fractured rocks, Ph.D. dissertation, 200 pp., Univ. of Ariz., Tucson, 1983. Kazemi, H., Pressure transient analysis of naturally fractured reservoirs with uniform fracture distribution, Trans. Soc. Pet. Eng. AIME, 246.451462, 1969. Kazemi, H., M. S. Seth, and G . W. Thomas, The interpretation of interference tests in naturally fractured reservoirs with uniform fracture distribution, Trans. Soc. Pet. Eng. AIME, 246, 463472, 1969. KuCuk, F., and W. K. Sawyer, Transient flow in naturally fractured reservoirs and its application to Devonian gas shales, Pap. SPE 9397, Soc. of Pet. Eng. of AIME, Dallas, Tex., 1980. Mavor, M. J., and H. Cinco L., Transient pressure behavior of naturally fractured reservoirs, Pap. SPE 7977, Soc. of Pet. Engineers of AIME, Dallas, Tex., 1979. Moench, A. F., and A. Ogata, Analysis of constant discharge wells by numerical inversion of Laplace transform solutions, in Groundwater Hydraulics, Water Resovr. Monogr. Ser., vol. 9, edited by J. Rosenshein and G. D. Bennett, pp. 146-170, AGU, Washington, D. c., 1984. Najurieta, H. L., A theory for pressure transient analysis in naturally fractured reservoirs, Trans. Soc. Pet. Eng. AIME, 269, 1241-1250, 1980. Neuzill, C. E., On conducting the modified "slug" test in tight formations, Water Resow. Res., 18(2), 439-441, 1982. Neuzil, C. E., and J. V. Tracy, Flow through fractures, Water Resour. Res., 17(1), 191-199, 1981. Papadopulos, I. S., Drawdown distribution around a large-diameter well, paper presented at the National Symposium on Groundwater Hydrology, Am. Water Resour. Assoc., San Francisco, Calif., 1%7. Papadopulos, I. S., and H. H. Cooper, Jr., Drawdown in a well of large diameter, Water Resow. Res., 3(1), 241-244, 1967. Ramey, H. J., Jr., and R. G. Agarwal, Annulus unloading rates as influenced by wellbore storage and skin effect, Trans. Soc. Pet. Eng. AIME, 253,453-462, I972 Sandal, H. M, J. W. Williamson, R. N. Horne, and H. J. Ramey, Jr., Interference testing with wellbore storage and skin at the produced well, Pap. SPE 7454, Soc of Pet. Eng. of AIME, Dallas, Tex, 1978. Scott, R. B., R. W. Spengler, S. Diehl, A. R. Lappin, and M. P. Chornack, Geological character of tuffs in the unsaturated wne at Yucca Mountain, southern Nevada, in Role of the Unsaturated Zone in Radioactive and Hazmdow Waste Disposal, edited by J . W. Mercer, et al., Ann Arbor Science Publishers, Ann Arbor, Mich., 1983. Snow, D. T., Anisotropic permeability of fractured media, Water Resour. Res., 5(6), 1273-1289,1969. Spengler, R. W., D. C. Muller, and R. 9. Livermore, Preliminary

..

846

MOENCH: DOUBLE-POROSITY MODUS FOR FISSURED GROUNDWATER RESERVOIR Warren, J. E., and P. J. Root, The behavior of naturally fractured ., reservoirs, Trans. Soc. Pet. Eng. AIME, 228,3(3), 245-255, 1963. Winograd, I. J., and W. Thordarson, Hydrogeologic and hydrochemical framework, south-central Great Basin, NevadaCalifornia, with special reference to the Nevada Test Site, U.S. ' Geol. Surv. Proj. Pap. 7 1 2 4 , 1 2 6 pp., 1975. I Witherspoon, P. A,, J. S. Y. Wang, K. Iwai, and J. E. Gale, Validity of the cubic law for fluid flow in a deformable rock fracture, Wolet Resour. Res., 16(6), 1016-1024, 1980. Young, R. A., Water supply for the nuclear rocket development station at the U.S. Atomic Energy Commission's Nevada Test Site, U.S. Geol. Suro. Wuter Supply Pap. 1938, 19 pp., 1972.

report on the geology and geophysics of drill hole UE25a-I, Yucca Moun~ain,Nevada Test Site, U.S. Geol. Surv. Open File Rep, 791244.43 pp., 1979. Stehfest, H., Numerical inversion of Laplace transforms, Commun. A C M , 13(1), 47-49, 1970. Streltsova, T. D., Hydrodynamics of groundwater flow in a fractured formation, Water Resour. Res., 12(3), 405-414, 1976. Streltsova, T. D., Well pressure behavior of a naturally fractured reservoir, Soc. Pet. Eny. J . , 23(5), 769-780, 1983. Tewley. J. D., Geologic characteristics of a portion of the Salton Sea geothermal field, Rep. UCRL-52267. 51 pp., Lawrence Livermore Lab.. Livermore, Calif., 1977. Theis, C. V., The relation between the lowering of the piezometric surface and the rate and duration of the discharge of a well using ground-water storage, Eos Trans. AGU, 16, 519-524, 1935. Tsang. Y. W.. and P. A. Witherspoon, Hydromechanical behavior of a deformable rock fracture subject to normal stress, J. Geophys. Res., 86(B10), 9287-9298, 1981. van der Kamp, G., and J. E. Gale, Theory of earth tide and barometric effects in porous formations with compressible grains, Water Resour. Res.. IY(2), 538-544. 1983.

A. F. ~ ~ U,S,D~~~~~~~~~ the interior, Geological Survey, ~ ~ of ~ h , Water Resources Division, 345 Middlefield Road. MS-496. Menlo park, CA 94025, (Received November 7, 1983; revised February 21. 1984; accepted March 5, 1984.)