SPE 71581
Interference Testing with Horizontal Observation Wells
M. Al-Khamis, SPE, and E. Ozkan, SPE, Colorado School of Mines, and R. Raghavan, SPE, Phillips Petroleum Co.
Copyright 2001, Society of Petroleum Engineers Inc.
This paper was prepared for presentation at the 2001 SPE Annual Technical Conference and
Exhibition held in New Orleans, Louisiana, 30 September–3 October 2001.
This paper was selected for presentation by an SPE Program Committee following review of
information contained in an abstract submitted by the author(s). Contents of the paper, as
presented, have not been reviewed by the Society of Petroleum Engineers and are subject to
correction by the author(s). The material, as presented, does not necessarily reflect any
position of the Society of Petroleum Engineers, its officers, or members. Papers presented at
SPE meetings are subject to publication review by Editorial Committees of the Society of
Petroleum Engineers. Electronic reproduction, distribution, or storage of any part of this paper
for commercial purposes without the written consent of the Society of Petroleum Engineers is
prohibited. Permission to reproduce in print is restricted to an abstract of not more than 300
words; illustrations may not be copied. The abstract must contain conspicuous
acknowledgment of where and by whom the paper was presented. Write Librarian, SPE, P.O.
Box 833836, Richardson, TX 75083-3836, U.S.A., fax 01-972-952-9435.
Abstract
Interference and pulse tests at horizontal wells introduce
features and issues that require specific procedures for
evaluation and analysis. In addition to reservoir properties, it
becomes necessary to account for three-dimensional pressure
distributions, as horizontal wells tend to be rather long. The
objective of this paper is to discuss the effect of the existence
of a horizontal observation well on the analysis of interference
tests.
We use a semi-analytical model for interference testing
with a horizontal observation well. The conductivity of the
horizontal wells may be infinite or finite. We discuss the flow
regimes in the observation well. The effect of anisotropy is
shown to significantly affect the interference between two
horizontal wells.
In many practical situations, it becomes necessary to
evaluate such tests by numerical means. At this time, no
simple method exists to calibrate observation well responses
computed by numerical simulators. The results of this study
should provide for an accurate measure of pressure responses
to calibrate numerical simulators.
Introduction
Analysis of horizontal well responses is considerably more
complex than that for their vertical counterparts.1 Similarly,
interference between two horizontal wells may create complex
pressure transients that may not be correlated by the
interference between two vertical wells.2 The complexity of
the horizontal well interference responses significantly
increases when the reservoir becomes more heterogeneous. In
fact, considering the long reach of horizontal wells,
homogeneous reservoir assumption may not be realistic for
horizontal interference test analysis. Therefore, in many
practical situations, horizontal well interference tests need to
be evaluated by using numerical simulators. Considering the
inherent problems of modeling horizontal wells in numerical
simulators, especially for transient analysis purposes, and the
fact that the details of flow in and around the horizontal
observation well also needs to be captured, the numerical
models used for horizontal well interference tests have to be
sensitively calibrated. Analytical models required to calibrate
numerical simulators for horizontal well interference testing
have not been reported until now.
The objective of this work is to present a semi-analytical
model that can be used to investigate the fundamental
characteristics of horizontal interference well responses and to
provide a means to calibrate numerical simulators. The model
consists of the superposition of the responses of two finite-
conductivity horizontal wells. Fluid withdrawal (production)
from one of the horizontal wells is considered at a constant
rate. The second horizontal well (observation well) is shut in
at the heel (for simplicity, we do not consider the effect of
wellbore storage in this work) but fluid may enter (production)
and leave (injection) this well along its length with total
production being equal to total injection. Both pulsing and
interference wells may have an arbitrary skin distribution
along their lengths. We have developed the solution presented
in this work by using the finite-conductivity horizontal well
model discussed in Refs. 3 and 4. The model provides the
pressure drawdown at the heel ends of the pulsing and
observation wells, as well as the flux and pressure
distributions along both horizontal wells. Therefore, the model
can be used to investigate the effect of interference not only at
the pressure measurement points (the heels of the wells) but
also to understand the local changes in and around the wells.
Below, we first introduce the physical system considered
and present the semi-analytical model. We, then, consider
some example cases to discuss the responses of both active
and interference wells. We investigate the effects of
anisotropy, skin, and finite wellbore conductivity. Finally, we
highlight the error that might be caused by ignoring the effect
of the length of a horizontal interference well (that is, treating
the observation well as if it were a vertical well). We also
compare the horizontal well interference test results with those
for the situation where both wells are vertical.
2 MOHAMMED AL-KHAMIS, ERDAL OZKAN, AND RAJAGOPAL RAGHAVAN
Model
As we mentioned in the Introduction, one of the main
objectives of this study is to develop a semi-analytical model
that can be used to investigate the fundamental features of
interference testing between two horizontal wells and to
calibrate numerical simulators. In this section, we discuss the
physical model considered and present the development of the
semi-analytical model. It should be noted that because of the
length of the derivations, we will only present the important
steps. The details of the derivations can be found in Ref. 5.
Physical Model. To develop the semi-analytical model, we
assume an anisotropic but homogeneous reservoir of
thickness, h. The flow is single phase and isothermal. We
consider a fluid of constant compressibility, c, and viscosity,
µ. Although the derivations presented here are for liquid flow,
they can be readily extended to single phase gas flow by the
pseudopressure concept. The reservoir pressure is uniform
initially and equal to pi.
Fig. 1 shows a sketch of the system considered in this
paper. We use the convention that Well 1 is the pulsing well
and Well 2 is the observation well. Fluid enters and leaves
both wells along their length. At a point x, the flow rate in
Well i, qhci, is given by the following relation:
Li + x wi
qhci ( x , t ) = ∫ qhi ( x' , t )dx' , (1)
x 6.328 × 10 −3 k
where q hi (x , t ) is the flux entering or leaving the well at point
x and time t (rate per unit length). A positive value for flux,
qhi, indicates influx (production) whereas a negative value
corresponds to outflux (fluid leaving the well or injection).
Note that the flux distribution along a well is a function of
time.
At the heel of the pulsing well (xw1), we set
qhci (x w1 , t ) = q = cons tan t production rate . Because there is
no production from the observation well and we assume that
the wells are produced or shut-in at the sandface (for
simplicity, we do not consider the effect of wellbore storage),
we impose the condition at the heel of the observation well
that q hci (x w 2 , t ) = 0 . The toes of the both wells (x = Lhi ) are
sealed.
Note that because there is no production from the
observation well, flux distribution along this well should be
induced by the pressure gradients caused by the production of
the active well. Specifically, if there is a pressure gradient
along the sandface of the observation well, the fluid will enter
the wellbore at the higher pressure points and travel toward
the lower pressure points (because the wellbore has higher
conductivity than the reservoir) where it will exit the wellbore.
By the requirement that the cumulative flux along the
observation well must be zero, the influx will be equal to the
outflux along the observation well. Therefore, until the radius
of investigation for the pulsing well reaches the observation
well, the flux distribution along the observation well should be
uniform and zero. After the radius of investigation of the
SPE 71581
pulsing well reaches the observation well, the far end of the
observation well should start producing fluid and the near end
should inject the fluid back into the reservoir. This indicates
that the observation well serves as a high conductivity path for
the production from the pulsing well. Because in practice, the
observation well will not be very close to the active well, the
contribution of the high conductivity flow path provided by
the observation well may not be noticeable at the active well.
(Note that the contribution of the observation well should be
in the form of a reduction in the drawdown.) The pressure
responses measured at the heel of the observation well,
however, may be significantly affected by the fluid flow in the
observation well. This is the underlying physical phenomenon
of our semi-analytical model. Before presenting the semi-
analytical model, however, we briefly introduce the
dimensionless varibales.
Dimensionless Variables. For convenience and generality,
we present our results in terms of dimensionless variables in
this study. We define the dimensionless pressure and time by
the following expressions, respectively:
kh
p D (t D ) = [ pi − p (x D , y D , z D ,t D )] , (2)
141.2 qBµ
and
tD = t, (3)
φ ct µ l 2
for t in days. The term l in Eq. 3 is a characteristic length in
the system that we choose as the half-length of the active well
in this study; that is, l = Lh1 2 . The permeability k is the
harmonic average of the permeabilities in the x, y, and z
directions ( k = 3 k x k y k z ). The dimensionless distances in x,
y, and z directions are defined, respectively, by
x k
xD = , (4)
l kx
y k
yD = , (5)
l ky
and
z
zD = . (6)
h
We also define the dimensionless formation thickness,
dimensionless horizontal well length, and dimensionless
wellbore radius by the following expressions, respectively:
h k
hD = , (7)
l kz
SPE 71581 INTERFERENCE TESTING WITH HORIZONTAL OBSERVATION WELLS 3

Lh kz PD (x D , y D , z D , t D ) = PD1 (x D , y D , z D ; x wD1 , y wD1 , z wD1 ,t D )

LD = , (8) + PD 2 (x D , y D , z D ; x wD 2 , y wD 2 , z wD 2 ,t D )
2h k
and (14)
where, pDi indicates the contribution of Well i ( i = 1 or 2) to
rwe the dimensionless pressure at point (xD, yD, zD) and is given by
rwD = . (9)
h Eq. 13.
where rwe is the equivalent wellbore radius for an anisotropic To obtain the wellbore flow model, we follow the lines
reservoir.6 Finally, the dimensionless flux is defined by discussed in Ref. 4. We will first discuss the wellbore flow
equation for Well 1 (active well). The steady-state momentum
2qh l kx equation for single-phase isothermal flow in a horizontal
qhD = . (10) wellbore is given by:
qB k
dPh1
In our derivations, we use two dimensionless groups to = Efqhc
2
, (15)
correlate the effect of wellbore hydraulics when the frictional dx
pressure drop in the wellbore is significant.4 These are the
Reynolds number and dimensionless conductivity defined, where Ph1 stands for the pressure at some point in Well 1, f is
respectively, by the following relations: the fanning friction factor, and

ρ q hc (x ) ρ
N Re c ( x ) = 6.166 × 10 −2 (11) E = 9.117 x10 −13 . (16)
µ rw π 2 rw5
and By differentiating Eq.15 with respect to x, we obtain;
2
rw4
C hD = 7.395 × 1013 . (12) dPh1  2 df1 
k h Lh = E  qhc1 − 2 f1qhc1qh1  (17)
dx 
2
dx 
Note that the above definitions are used for both the active and
In Eq. 17, qhc1 is the flow rate in Well 1 (at point x) and qh1 is
observation wells. We use the subscript 1 with the above
the flux entering or leaving the well at the same point. In Eq.
definitions to indicate the property related to the active well
17, we have used the following relation between qhc1 and qh1:
and subscript 2 for the properties of the observation well.
 ∂qhc1 
Semi-Analytical Model. The semi-analytical model is based   = −qh1(x,t) (18)
 ∂x  ( x ,t )
on the superposition of two finite-conductivity horizontal well
solutions for the physical conditions discussed above. The Integrating Eq. 17, and using the boundary condition at the
development of the finite-conductivity horizontal well solution heel of Well 1 given by
has been discussed in Refs. 3 and 4 and involves coupling of
 dPh1 
the wellbore and reservoir flow equations. The reservoir flow   = Ef t1 q 2 B 2 , (19)
equation [dimensionless pressure at a point (xD, yD, zD)] for a  dx  x = xw1
single horizontal well of length, Lh, located at (xwD, ywD, zwD) in
an infinite slab reservoir of thickness, h, is obtained by using where q is the surface production rate of Well 1 (constant) and
the method of sources and sinks and is given by ft1 is the friction factor at the heel of this well, we obtain the
following expression:
k2
PD (x D , y D , z D , t D ) =
1 x 
dPh1 2 df1 
4 kxky = Ef t1q 2 B 2 + E ∫  qhc1 '
-2 f1qhc1qh1 dx' (20)
dx x w1  dx 
k
2 LD h D
tD kx  (x − x wD − α )2  Integrating Eq. 20 once more, we have
∫ ∫ q hD ( α ,τ ) exp  − D  dα
0 0  4( t D − τ ) 
Ph1 ( x , t ) − Pwf 1 = Ef t1q 2 B 2 ( x − x w1 )
 ( y − y wD ) 2 
exp  − D  x x'
 2 df1 
 4( t D − τ )  +E ∫ ∫  qhc1 -2 f1qhc1qh1 dx" dx'
x w1 x w1  dx" 
 ∞  n 2π 2 ( t D − τ )   dτ
1 + 2 ∑ exp  −  cos nπz D cos nπz wD  (21)
 ( t D − τ )
2
 n =1
 hD 

(13)
To obtain the dimensionless pressure because of two
horizontal wells, we superpose Eq. 13 as follows:
4 MOHAMMED AL-KHAMIS, ERDAL OZKAN, AND RAJAGOPAL RAGHAVAN SPE 71581

In terms of dimensionless variables, Eq. 21 can be written as: Development of the coupled flow equation for the
observation well follows the same lines as above except,

 because the observation well is shut-in, the boundary
PwD 1 ( t D ) − PhD1 ( x D , t D ) =
N Re t1 f t1 π  2( x D − x wD 1 ) condition in Eq. 19 must be replaced by
16 C hD1  k
 L D1 h D  dPh1 
 kx   =0 (27)
 dx  x = x w1

 This yields the following expression for the observation well:
xD  D1
xD ' 
q hD1 dx" dx 'D 
1
− ∫ ∫  PwD 2 ( t D ) − PhD 2 ( x D , t D ) =
k x wD1 x wD1  N Re t1 f t1 
L D1 h D  xD xD'
kx  π '
− ∫ ∫ D 2 q hD 2 dx" dx D
k x wD 2 x wD 2
(22) 16C hD 2 L D1 h D
kx
where we have used
 df1   N Re 1 df  (28)
N Re 1 1 k
  − qhD1 = −  + 2  qhD1 where
N Re t1 f 1  dx D  2l kx  f1 dN Re 1 
 df 2 
(23) D 2 = (N Re 2 )2   + 2 N Re 2 f 2 (29)
and defined  dN Re 2 

 df1  and phD2 is given by Eq. 25 with k = 2.

D1 = (N Re 1 )2   + 2 N Re1 f1 .
 (24)
 dN Re1 
The terms pwD1 and PhD1(xD,tD) in Eq. 22 are the dimensionless
wellbore pressures measured at the heel of the active well and
at some point xD, respectively. Note that by the continuity of
pressure, PhD1(xD,tD) is equal to the sandface pressure and from
Eq. 14, can be written as
PhDk (x Dk , y wDk , z wDk , t D ) =
PD1k (x Dk , y wDk , z wDk + rwDk ; x wD1 , y wD1 , z wD1 , t D )
+ PD 2k (x Dk , y wDk , z wDk + rwDk ; x wD 2 , y wD 2 , z wD 2 , t D )
+ q hDk (x Dk , t D )S k (x Dk )
(25)
for k = 1 and xDk is along the horizontal well. The Sk term in
Eq. 25 is the mechanical skin defined by7
kh
∆p k ,skin ( x k ,t )
141.2qBµ
S k ( x Dk ) = . (26)
LhDk h D k k x qhDk ( x Dk ,t D )
Remember that the terms pDik in Eq. 25 are given by Eq.
13. Again by the continuity of flux along the surface of the
well, qhDi terms (i = 1) in Eqs. 22 and 25 are the same.
Therefore, Eqs. 22 and 25 represent the coupled wellbore-
reservoir flow model for the active well. As shown in Refs. 4
and 5, the coupled flow equation may be discretized along the
horizontal well and solved for the wellbore pressure, pwD1, and
the dimensionless flux, qhD1. Note, however, that in our two-
Eqs. 22 and 28 constitute the coupled flow equations for
finite-conductivity active and observation wells. Note here
that, as our results also show, we do not normally expect
considerable frictional pressure drop in the observation well
because the flow rates in this well should be small. Because
one of the objectives of this study is to investigate the validity
of our common expectations and assumptions, here we model
the observation well to take into account the finite-
conductivity effects.
To solve Eqs. 22 and 28, we discretize both horizontal
wells into M segments and evaluate Eqs. 22 and 28 at the
center of each segment i (xDki). The discretization of the flow
equations follow the same lines as in Ref. 4 and will not be
discussed here. The discretized flow equations yield 2M
expressions in 2M+2 unknowns, pwD1, pwD2, qhD1, and qhD2.
Two additional expressions can be obtained from the
condition that the sum of the fluxes along the active well must
be equal to the production rate and sum along the observation
well, must be zero. These conditions can be expressed,
respectively, as follows:
M M
∑ q hD1i = (30)
i =1 k
LhD1 h D
kx
and
M
∑ qhD 2 i = 0 (31)
i =1

well problem, the computation of the pD2k term in Eq. 25
requires the flux distribution for Well 2. Therefore, to be able
to evaluate the coupled flow equation for the active well, we
need to know the flux distribution for the observation well.
This can be accomplished by developing the coupled flow
equation for the observation well.
where qhDki is the flux of Well k at the center of the ith segment
(xDki).
The discretized forms of Eqs. 22 and 28 with Eqs. 30 and
31 provide 2M+2 equations in 2M+2 unknowns. This system
of equations needs to be solved by an iteration process (we
have used Newton-Raphson method) because Eqs. 22 and 28
SPE 71581 INTERFERENCE TESTING WITH HORIZONTAL OBSERVATION WELLS
are non-linear (D1 and D2 are functions of qhD1 and qhD2,
respectively). The solution procedure is similar to that
discussed in Ref. 4 and the additional details can be found in
Ref. 6.
This completes the discussion of our semi-analytical
model. In the following section, we will consider the results
generated by this model to investigate the fundamental
features of interference testing with horizontal wells. Before
passing, however, we present some data generated by our
model. This data should be useful to calibrate reservoir
simulators.
Table 1 shows the data used as our base case. In most of
our discussions in this paper, we only change the relative
location of the observation well with respect to the pulsing
well. Also, for simplicity, we always assume that the
properties of Wells 1 and 2 are the same. Table 2 presents the
pressure and logarithmic derivative responses computed at the
active and observation wells for the base case. Here the heel of
the observation well is located halfway the length of the active
well and the two wells has a separation of one-and-a-half
times the length of the wells in the normal (y) direction. Table
3 is considers a case of stacked laterals. The active and
observation wells in this case are aligned in the vertical plane
and the vertical distance between the wells is ∆z = 33.3 ft.
Tables 2 and 3 may help those who would like to reproduce
our results analytically or calibrate their numerical models.
Results and Discussion
Here we consider some example cases of interference testing
with horizontal wells. We first highlight the significance of
modeling the observation well as a horizontal well. We, then,
consider the effect of the separation between the pulsing and
observation wells in the x (axial) and y (normal) directions.
General discussion of the transient pressure characteristics
follows. We, finally, consider the effect of finite wellbore
conductivity.
In order to emphasize the importance of rigorously
modeling the active and observation wells, in Fig. 2, we show
the flux distributions along the active (Well 1) and observation
(Well 2) wells for the base case given in Table 1. In this
example, the distance between the heels of the wells in the x
direction is ∆xw = 1000 ft and the separation of the wells in the
normal direction is ∆yw = 3000 ft. As expected, Fig, 2 shows
that the flux distribution is uniform at early times for the
active well and it takes a u-shape at late times. This is the
characteristic behavior for infinite-conductivity horizontal
wells. (Note that the wells in the examples discussed here are
all infinite-conductivity except for the section where we
discuss the effect of finite-wellbore conductivity). The
observation well flux distribution, on the other hand, is zero at
early times because the radius of influence of the active well
does not cover the observation well at these times. When the
effect of production is felt at the observation well location (at
late times), a different flux profile develops. Approximately
half of the horizontal well segments (Segments 6-10) have
positive flux (production) and the other half (Segments 1-5)
5
have negative flux (injection). That means, the portions of the
observation well far away from the active well produces the
fluid from the reservoir and transfers it to the sections closer to
the active well where the fluid exits the observation well to
converge toward the active well. This observation is consistent
with our physical expectations discussed earlier. Note that to
capture the details of the flux distributions (or flow in the near
vicinity of the wellbores) shown in Fig. 2, rigorous models
such as the one presented here should be used.
Fig. 3 shows the effect of the axial separation (∆xw) of the
active and observation wells at t = 128 days. There is
practically no effect of the axial distance between the two
wells on the active well responses and a small effect on the
observation well responses. The separation in the normal
direction (∆yw), however, has a more significant influence.
This result should be attributed to the fact that the y-direction
permeability is twice as large as the x-direction permeability.
Fig. 4 displays the influence of vertical distance (∆z) for
stacked laterals (active and observation wells are aligned in
the vertical plane). In this case, the observation well pressures
are affected slightly by the distance between the wells. More
significant effect, however, is on active well responses. Fig. 4
indicates that the drawdown at the active well increase as the
distance between the wells increases. This is a result of the
fact that when the stacked laterals are closer to each other, the
observation well helps the production of the active well (by
providing a high conductivity flow path). This statement is
supported by the flux distributions shown in Fig. 5. The tips of
the observation well withdraw from the reservoir and the
central segments inject the fluid into the reservoir toward the
active well.
We, now, consider some examples of transient pressure
and derivative plots to discuss the characteristics of active and
observation well responses. We begin with the base case
responses shown in Fig. 6. As expected from the discussion of
the flux distribution and drawdown at the active and
observation wells (Figs. 2-5), the active well responses are not
affected by the existence of a horizontal interference well.
Especially the derivative responses of the active well clearly
display the characteristic flow regimes (early-time radial flow,
intermediate- time linear flow, and late-time pseudoradial
flow) of horizontal wells. We have seen in Fig. 2 that,
although small, some influx (production) and outflux
(injection) takes place along the observation well. Fig. 6,
however, indicates that the pressure responses measured at the
observation well display characteristics more in line with
radial or pseudoradial flow. We will investigate this
observation in further detail later.
Fig. 7 considers a different orientation of the wells. This is
a stacked laterals case and the vertical distance between the
wells is 33.3 ft. Similar to Fig. 6, the active well responses do
not display any signature of interference. The observation well
responses, however, becomes significant much earlier
compared to the case shown in Fig. 6. More importantly, in
this case the derivative responses of the observation well
6 MOHAMMED AL-KHAMIS, ERDAL OZKAN, AND RAJAGOPAL RAGHAVAN
display an almost linear flow regime (the intermediate-time
data follow a one-half slope straight line).
We examine the interference test responses of opposing
dual laterals in Figs. 8 and 9. In Fig. 8, the distance between
the laterals is 1000 ft and in Fig. 9 the distance is 2000 ft. The
characteristics of the active and observation well responses are
similar to those shown in Fig. 6. As expected, it can be seen
from Figs. 8 and 9 that, as the distance between the wells
increases, the observation well responses are delayed. The
general characteristics of the responses remain the same.
We have also investigated the effect of skin distribution
around the active and observation wells. Fig. 10 shows three
cases for the active well: no-skin, uniform skin (S=0.5) and
conical skin ( 0.05 ≤ S ≤ 0.5 ). As discussed in Ref. 7, the
conical skin distribution causes some distortion of the early-
and intermediate-time flow regimes. When we considered the
observation well responses, however, (the results are not
shown) we did not see any noticeable effect of the skin
distributions in both wells. This result is explained by the
small magnitude of the flux on the surface of the observation
well.
In Figs. 11 and 12, we investigate the possibility of
interpreting the observation well responses as if the well was
vertical (practically, in this case the observation well is point
in the reservoir). In Fig. 11, the actual horizontal wells are
aligned in the horizontal plane and in Fig. 12, they are axially
displaced. It is interesting to note that when the horizontal
observation well is aligned with the active well, its responses
can be closely correlated by those of a vertical well located at
the heel of the actual horizontal observation well. When the
horizontal observation well is displaced in the axial direction,
however, no correlation is possible with the vertical
observation well case. This result can be explained by the fact
that the axial displacement of the horizontal well causes more
distortion of the pseudoradial flow characteristics.
As we briefly mentioned before, the observation well
responses display characteristics similar to those of radial or
pseudoradial flow. In Figs. 13 and 14, we compare the
observation well responses with the exponential integral
solution for the base case and the opposing dual laterals. In
both figures, we consider the radial distances measured from
the heel of the active well to the heel of the observation well
and to the mid-point of the observation well. It is clear that,
although the characteristics are similar, no correlation between
the interference and exponential integral solutions is possible.
For completeness, in Fig. 15, we show the derivative
responses for the opposing dual laterals case. This figure
indicates that at late times, the horizontal interference well
responses display radial flow characteristics as the exponential
integral solution. At earlier times, however, no resemblance
can be found with the known flow regimes.
Before closing this section, we also comment on the effect
of finite wellbore conductivity. In this study, we have also
considered cases where the wellbore conductivity was finite.
In all the cases we examined, we found that the effect of
finite-conductivity was only noticeable in active well
SPE 71581
responses. No noticeable effect was observed in the
observation well responses even when the two laterals were
reasonably close. This indicates that the distortion of the
equipotential lines because of finite conductivity is only
effective in the near vicinity of the wellbore and does not
reach into the reservoir. Also, because the flow rates in the
observation well were normally small, no frictional pressure
drop can be expected.
Conclusions
In this study, we developed a semi-analytical model and
investigated the general characteristics of interference test
responses with horizontal wells. The semi-analytical model
presented is a rigorous model for reasonably homogenous
reservoirs and can be used to analyze interference well tests
with horizontal wells. The model can also be used to calibrate
the numerical simulators. The results generated by the semi-
analytical model indicate that the vertical interference well
assumption is not always valid for interference testing
between two horizontal wells. Similarly, the horizontal
observation well responses do not follow the exponential
integral solution. These results indicate that the interference
tests between two horizontal wells should be analyzed by
using either semi-analytical models such as the one presented
in this study or by reservoir simulators.
Nomenclature
C hD = dimensionless horizontal well conductivity
c t = total compressibility, psi-1
h = formation thickness, ft
k = permeability, md
p = pressure, psi
q = production rate, stb/d
q DND = dimensionless flow rate constant
re = reservoir radius, ft
r′w = effective wellbore radius, ft
t = time, hr
Z = gas compressibility factor
S = Skin
Greek symbols
φ = porosity
γ = Euler’s constant (=0.5772)
µ = viscosity, cp
Subscripts and superscripts
D = dimensionless
f = fracture
i = initial
w = wellbore
wf = flowing wellbore
ws = shut-in well
SPE 71581 INTERFERENCE TESTING WITH HORIZONTAL OBSERVATION WELLS 7

References Table 1. Base-case data

1. Ozkan, E.: “Analysis of Horizontal Well Responses: q, STBD 2000
Contemporary vs. Conventional,” paper SPE 52199 presented at Lh, ft 2000
the 1999 SPE Mid-Continent Operations Symposium, Oklahoma h, ft 100
City, OK, March 28-31, 1999.
2. Umnuayponwiwat, S., Ozkan, E., and Raghavan, R.: “Pressure kx, md 50
Transient Behavior and Inflow Performance of Multiple Wells ky, md 100
in Closed Systems,” paper SPE 62988 presented at the 2000 kz, md 25
SPE Annual Technical Conference and Exhibition, Dallas, TX, rw, ft 0.165
Oct. 1-4, 2000. ∆xw, ft 1000
3. Ozkan, E., Saric, C, Haciislamoglu, M., and Raghavan, R.:
“Effect of Conductivity on Horizontal-Well Pressure-Behavior,”
∆yw, ft 3000
SPE Advanced Technology Series, Vol 3, No. 1, p.85, 1995. ∆zw, ft 50
4. Ozkan, E., Sarica, C., and Haciislamoglu, M: “The Influence of
Pressure Drop Along the Wellbore on Horizontal Well
Productivity,” SPEJ (Sept. 1999) 180. Table 2. Pressure and derivative data for the base case
5. Al-Khamis, M.: Analysis of Interference Tests with Advanced tD pwD1 p'wD1 pwD2 p'wD2
Well Geometries, Ph.D. Dissertation, Colorado School of Mines,
1.0E-05 0.1545 0.0250 0.0000 0.0000
Golden, CO (2002).
6. Peaceman, D.: “Interpretation of Well-Block Pressures in 1.0E-04 0.2120 0.0250 0.0000 0.0000
Numerical Reservoir Simulation with Nonsquare Grid Blocks 1.0E-03 0.2696 0.0253 0.0000 0.0000
and Anisotropic Permeability,” SPEJ (June 1983) 531. 1.0E-02 0.3553 0.0594 0.0000 0.0000
7. Ozkan, E. and Raghavan, R.: “Estimation of Formation Damage 1.0E-01 0.5924 0.1595 0.0000 0.0000
in Horizontal Wells,” paper SPE 37511 presented at the 1.0E+00 1.1246 0.2962 0.0396 0.0820
Production Operations Symposium, Oklahoma City, OK, March
1.0E+01 1.8782 0.3450 0.5180 0.3019
9-11, 1997.
1.0E+02 2.6844 0.3523 1.2841 0.3480
SI Metric Conversion Factors 1.0E+03 3.4959 0.3591 2.0931 0.3531
1.0E+04 4.3532 0.5249 2.9072 0.3545
bbl x 1.589 873 E-01 = m3
cp x 1.0* E-03 = Pa*s
ft x 3.048* E-01 = m Table 3. Pressure and derivative data for stacked laterals
ft3 x 2.831 685 E-02 = m3 tD pwD1 p'wD1 pwD2 p'wD2
in. x 2.54* E+00 = cm 1.0E-05 0.1545 0.0250 0.0000 0.0000
lbf x 4.448 222 E+00= N 1.0E-04 0.2120 0.0250 0.0000 0.0001
lbm x 4.535 924 E-01 = kg 1.0E-03 0.2705 0.0277 0.0124 0.0146
mD x 9.869 233 E-04 = µm2 1.0E-02 0.3623 0.0596 0.0882 0.0592
psi x 6.894 757 E+00 = kPa 1.0E-01 0.5982 0.1580 0.3242 0.1580
1.0E+00 1.1246 0.2939 0.8507 0.2941
*Conversion factor is exact. 1.0E+01 1.8784 0.3461 1.6045 0.3463
1.0E+02 2.6857 0.3525 2.4119 0.3525
1.0E+03 3.4973 0.3590 3.2252 0.3538
1.0E+04 4.3543 0.5247 4.0415 0.3597
8 MOHAMMED AL-KHAMIS, ERDAL OZKAN, AND RAJAGOPAL RAGHAVAN SPE 71581

350
Z t= 128 days
zw2

∆ pw1 and ∆ pw2, psi

310 Lh= 2000 ft
HW 2 Well 1
270
zw1 Lh
HW 1
230
Lh
Well 2
xw1 xw2 190 ∆z
yw1 (0,0, X
yw2 150
Y 0 40 80 120
∆ z, ft
Fig. 4 – Change in the drawdown at the active and observation
Fig. 1 – Schematic of the system considered in this study.
wells as a function of vertical separation. t=128 days.
1.8 1.5
Well 1 Well 1
1.5 tD=1E-5 , t D=1E2
tD=1E-5 , t D=1E2 1.2
q hD1 and q hD2

1.2

0.9
qhD1 and q hD2 0.9
Lh= 2000 ft
0.6 Lh = 2000 ft 0.6 ∆ z= 33.3 ft
∆ xw= 1000 ft Well 2, t D=1E2
0.3 ∆ yw= 3000 ft 0.3 Well 2, t D=1E2

0.0
0.0
1 2 3 4 5 6 7 8 9 10
-0.3 1 2 3 4 5 6 7 8 9 10
Well Segment, i -0.3
Well Segment, i
Fig. 2 – Flux distributions for the active (Well 1) and observation Fig. 5 – Flux distributions for vertically parallel dual laterals.
wells (Well 2).
300 10.00
PwD1
Well 1 P'wD1
250
pwD and p'wD

PwD2
t= 128 days
∆pw1 and ∆pw2 , psi

∆xw 1.00 P'wD2

Lh= 2000 ft
200
∆yw
150 Well 2 ∆yw = 3000 ft 0.10
Well Location
100

∆yw = 6000 ft 0.01

50
1.E-05 1.E-03 1.E-01 1.E+01 1.E+03
0 tD
0 500 1000 1500 2000 2500 3000
Fig. 6 – Dimensionless pressure and derivative responses for the
∆ xw, ft base case.
Fig. 3 – Change in the drawdown measured at the active and
observation wells as a function of the axial separation of the
wells. t=128 days.
SPE 71581 INTERFERENCE TESTING WITH HORIZONTAL OBSERVATION WELLS 9

10.00 10.00
PwD1 Well Location
P'wD1
PwD2
P'wD2

pwD and p'wD

pwD and p'wD

1.00 1.00

0.10 0.10
Well Location PwD1
P'wD1
h PwD2
P'wD2
0.01 0.01
1.E-05 1.E-03 1.E-01 1.E+01 1.E+03 1.E-05 1.E-03 1.E-01 1.E+01 1.E+03
tD tD
Fig. 7 – Dimensionless pressure and derivative responses of Fig. 9 – Dimensionless pressure and derivative responses for
stacked laterals in interference testing. ∆ xw=2000 ft).
opposing dual laterals (∆

10.00 10.00
PwD1 Well Location Well Location
P'wD1
PwD2 pwD1
P'wD2
pwD and p'wD

pwD and p'wD

1.00 1.00

p'wD1
0.10 0.10
PwD1
PwD1(Skin= 0.5)
PwD1(Skin= 0.5-0.05)

0.01 0.01
1.E-05 1.E-03 1.E-01 1.E+01 1.E+03 1.E-05 1.E-03 1.E-01 1.E+01 1.E+03
tD tD
Fig. 8 – Dimensionless pressure and derivative responses for Fig. 10 – Skin effect on active well response for the base case.
∆ xw=1000 ft).
opposing dual laterals (∆
10 MOHAMMED AL-KHAMIS, ERDAL OZKAN, AND RAJAGOPAL RAGHAVAN SPE 71581

10.00 100.00
H-V Well Location
H-H

10.00
1.00

pwD2
pwD2

1.00

H-H
0.10 Exp(rD=2.12)
0.10
Well Location Exp(rD=2.34)

0.01
0.01 1.E-01 1.E+01 1.E+03 1.E+05 1.E+07
0.1 1 10 100 1000
tD tD
Fig. 11 – Comparison of the dimensionless pressure responses Fig. 13 – Comparison of the observation well responses with
of horizontal and vertical observation wells. Active and exponential integral solution for the base case.
observation wells are aligned.

10.00 100.00
H-V Well Location
H-H

10.00
1.00
pwD2
pwD2

1.00

0.10
Well Location
0.10 H-H
Exp(rD= 1)
Exp(rD= 2)
Exp(rD= 3)

0.01
0.01
1.E-01 1.E+01 1.E+03 1.E+05 1.E+07
0.1 1 10 100 1000
tD tD
Fig. 12 – Comparison of the dimensionless pressure responses Fig. 14 – Comparison of the observation well responses with the
of horizontal and vertical observation wells. Active and ∆ xw = 1000
exponential integral solution. Opposing dual laterals (∆
observation wells are axially separated. ft).
SPE 71581 INTERFERENCE TESTING WITH HORIZONTAL OBSERVATION WELLS 11

100.00
Well Location

10.00
pwD2 and p'wD2

1.00

0.10 PwD(H-H)
P'wD(H-H)
PwD(Exp(rD= 2))
P'wD(Exp(rD= 2))

0.01
1.E-01 1.E+00 1.E+01 1.E+02 1.E+03
tD

Fig. 15 – Comparison of the observation well derivative responses

with the derivative of the exponential integral solution. Opposing
∆ xw = 1000 ft).
dual laterals (∆