You are on page 1of 10

SPE

SPE 23005

Optimization of Perforation Distribution for Horizontal Wells


M.J. Landman, BHP Research Melbourne Laboratories, and W.H. Goldthorpe,
BHP Petroleum
SPE Members

Copyright 1991, Society of Petroleum Engineers, Inc.

This paper was prepared for presentation at the SPE Asia-Pacific Conference held in Perth, Western Australia, 4-7 November 1991.

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. Permission to copy is restricted to an abstract of not more than 300 words. Illustrations may not be copied. The abstract should contain conspicuous acknowledgm'ent
of where and by whom the paper is presented. Write Librarian, SPE, P.O. Box 833836, Richardson, TX 75083-3836 U.S.A. Telex, 730989 SPEDAL.

Abstract literature, perhaps because of the relatively small number of


In a high permeability reservoir, significant pressure drop horizontal wells currently in production for which it is
can occur in a horizontal well, causing nonuniform inflow due perceived as relevant.
to the variation in pressure drawdown along the well. A
mathematical model is described for investigating how If the flowing pressure gradient within the horizontal
perforation distribution affects the performance of a wellbore is comparable to the drawdown in the well, both
horizontal well producing under steady state inflow. The need to be taken into account when designing the
model couples the Darcy flow into each perforation with the completion, and in particular the perforation distribution.
one-dimensional momentum equations for pipe flow. The This can occur in long highly productive wells, wells with
perforated well is treated like a pipe manifold with T- small diameter liners, and wells with high gas-oil ratios.
junctions representing the perforations along the wellbore. When horizontal wells are used to minimize water or gas
A number of examples are presented demonstrating how the coning, an inappropriate perforation distribution may
model may be used to determine the optimal perforation actually exacerbate the problem at the downstream end of the
distribution for different production requirements. well.

Introduction Dikken8 presented the first semi-analytical model to calculate


the production performance of a horizontal well with a
One of the distinguishing features of horizontal wells wellbore pressure gradient resulting from turbulent flow. A
compared to conventional wells is the relatively large well similar model was developed earlier in the context of water
length which penetrates the reservoir and is available for uptake by plant roots by Landsberg and Fowkes9 • Frictional
perforating. Performance of a horizontal well may therefore pressure drop calculations have since been added to reservoir
depend not only on the mechanism by which fluid enters the simulators in a variety of wayslO-l3, including incorporating
well, but also on the fluid flow within the horizontal section, models similar to Dikken's. The model of Islam and
particularly in high permeability reservoirs where relatively Chakmal l , although allowing for the effect of inflow through
small pressure drawdown occurs. perforations on the two-phase flow regime within the
wellbore, does not provide a means to investigate the impact
A considerable amount of analytical work has been published of different perforation distributions on well performance. A
on various aspects of horizontal well production, including number of studies and models have been presented on
transient flow models, stabilized inflow models and perforated completions in conventional oil and gas wells (for
productivity indices, and coning and cresting behavior l -7 . example Ichara l4 ), but there has not been any discussion on
Although these methods have provided insight into the the application of these to horizontal wells with wellbore
behavior of horizontal wells, they have generally been based pressure gradients.
on the assumption of the well being a line sink with a
uniform flux boundary condition at the well. Hence any This paper describes a mathematical model that couples the
impact of fluid flowing within the wellbore on the well's pressure-flowrate relationship for each of a specified number
inflow performance is neglected. This aspect of horizontal of perforations with a wellbore pressure drop model that treats
well modeling has not received much attention in the the perforated well as a pipe manifold containing T-junctions.
Flow is assumed to be single phase and turbulent within the
References and figures at end of paper. wellbore, although it would be possible to incorporate

567
2 OPI1MlZATION OF PERroRATION DISTRIBUIlON roR HORlZONfAL WELLS SPE23005

laminar flow or multiphase flow (e.g. bubble-induced essentially the same as a point sink model 16. 17 with wellbore
turbulencel l ) in the wellbore. The primary objective of the radius
model is to provide a tool for investigating the effect of
different perforation distributions on horizontal well (3)
performance. Thus. for a given production strategy, the
optimal distribution under steady state inflow can be
Note, however, that unlike previous studies, the sink
detennined.
positions ]Ci are considered here as variables, on which the
A set of basic optimization strategies are defined, and applied pressure and flowrate depend nonlinearly.
to both inImite conductivity (constant pressure) and finite
conductivity (with a wellbore pressure gradient) wells. In the Writing the values of pressure and flowrate along a well of N
infinite conductivity case, it is demonstrated how non- perforations as N-vecrors, equation (2) can be written
uniformly spaced perforations are required for various (4)
optimization problems. This results from higher flow rates
into end perforations due to partial penetration. For f"mite where the matrix A depends on the perforation size and
conductivity wells, a comparison is made with both the distribution. Therefore, given the pressure distribution along
approach of Dikken and with the case when an infinite the well, and a given perforation distribution, it becomes an
conductivity assumption is made. Examples of well NxN matrix inversion problem to find the flowrate per
optimization are presented, resulting in significant variation perforation:
in perforation density, particularly due to the variation in
(5)
drawdown along the well. A simple application of the
methodology to a well penetrating zones of different This is a discrete analogue of the problem of modeling a fully-
penneability is also included. open horizontal well with specified (e.g. constant) pressure
boundary conditions, which has only been solved
Throughout this paper the following conditions are assumed. numerically to datelS. Note, however, that parallel to the
The reservoir is infinite, isotropic and homogeneous (except well, the pressure is far from constant in the perforated case,
for the multiple zone model), with a constant pressure as there are local pressure minima at each of the perforations.
boundary at infinity. There is no contribution to inflow
pressure drawdown from skin due to formation damage. All Manifold Model for Pressure Drop In Well
fluid flow is single phase, with turbulent flow in the wellbore.
The model proposed allows for a finite conductivity well, due
The Perforation Model for the Reservoir to the frictional pressure drop of fluid flow within the well.
The pressure drop in a horizontal well is discussed by
A schematic diagram of the well model is shown in Figure 1. Dikken8 , who models an open-hole or slotted liner
The perforations are considered to be cylindrical, of length a, completion. The significance of well pressure drop has
and radius S. The influence of the wellbore and damage to the recently been quantified by comparing the dissipation of
surrounding formation is assumed to be negligible. An energy in the well and in the reservoir19 • In general, the
analytic formula for the flow into an infinite conductivity pressure drop along the well, I1P, is significant given a
cylinder (i.e., on which the boundary condition is that the sufficiently long well of small diameter, and sufficiently
pressure is specified) is unavailable. The pressure-flowrate small drawdownP d at the downstream end of the well, so that
relationship is known, however, in the limit of small lila. In J1P becomes comparable with Pd' This is not atypical for a
that case, the self induced pressure due to flow Q i into the perforated completion in a high penneability reservoir.
perforation itself is given by
Dikken assumes turbulent pipe flow in the well. The current
LQ.1oJ!) (I) work for high productivity wells shows that this assumption
2~k a 6\a is justified for essentially the entire length of the well.
where the reservoir pressure datum is taken to be zero "at Although Dikken's model allows for a variable specific
infinity", and the natural logarithm is used. This may be productivity index (PI), he concentrates on having a uniform
derived as the leading term, as Sla ~ 0, for either the average PI per unit length. The results which follow demonstrate that
pressure at a radius Ii around a finite length line source of end effects can be important, however, as the specific PI
uniform strength. or from the more sophisticated model where increases there.
the perforation is a large aspect ratio ellipsoid, on which
constant pressure is specified14• The pressure drop model considers the perforated well to be
The separation between perforations is sufficiently great so like a pipe manifold, with NT-junctions distributed along the
that at a given perforation i, every other perforation j has the wellbore of diameter D, and cross-sectional area A. The
influence of a point sink of strength Qj. Therefore the pressure drop in the wellbore across each perforation outlet is
phasing of the perforations need not be considered. The assumed negligible, as is the pressure drop within a
pressure at perforation i is therefore given by perforation. Estimates of these pressure drops (for single-
phase flow) show that they are much smaller than that due to

P; = L~_ 2Q.a IOg(!)+


4~ a
L.. ~Qj
. -x-
I]
J"'. J
(2)
turbulent flow in the pipe segments between perforations20 •

The pressure drop due to friction is derived from a single


phase pipe flow model, where
where perforation j is positioned at x=xj ' with Xl the f"rrst
perforation at the downstream end of the well. This is ap_ pI q2 (6)
aX- 2DA2

568
SPE23p05 M. J. LANDMAN AND W. H. GOlDIHORPE 3

The pressure drop due to a chmge in kinetic energy is judged to have converged when both the sum of the
neglected (it is estimated as being negligible compared to increments in Pi and Q i are less than a given small 1Olerance.
that from friction); the chmge in head due 10 well inclination This usually occurs in small number of iterations. due to a
is not included in (6). but can be easily incorporated in the geometric convergence rate.
model.
When an inf"mite conductivity wellbore is assumed. the
On discretizing (6). the pressure Pi at each perforation i along pressure vec10r is constant and equal 10 Pd' so that (12a) can
the well is be directly applied, and no iterations are required.
Ii =Pd (7a)
Optimization Strategies
- + 2D
P;+1 -P; Phq? I~
A2 l""i+1 -Xi
I (7b) Optimizing the perforation distribution involves a number of
competing factors. First, the reservoir produces most
P d is the downstream drawdown, which is taken as specified at efficiently when the perforations are placed as far apart as
position xl' Alternatively. the total flowrate could be possible. so that increasing the well length is desirable.
specified, although this requires a slightly more complex Second. the pressure drop due to fluid flow in the well
solution method, as described in Appendix A. increases with L. in turn decreasing the inflow towards the
upstream end of the well. Third. finite penetration increases
Equation (7b) expresses the pressure drop in successive pipe the specific productivity of the well at its ends.
segments i (i=l.N-l), in which the cumulative flowrate is
N
There are several optimization strategies which take into
account such fac1Ors. each of the general form: Given a well of
~=~~ 00 length L. which is to be perforated with N charges, what is the
j=i+1
optimal distribution of the charges to maximize the well's
The pressures are thus calculated from the perforation performance? The well's performance can be measured by
flowrates. working upstream from the first perforation. The several different quantities, depending on the physical
turbulent friction factor is taken ail° circumstances (reservoir heterogeneity. boundaries. aquifers
and gas-caps). In the simple geometry considered here, five
2.-+ 5'~~)J2
=0.25[101~3.7D prototype optimization problems have been identified:
h (9)
Rei
PROBLEM lA: Maximize the total production from the well
with pipe rouglmess 1C and local Reynolds number N
Q= ~Q; (13)
Re. = P Dqi (10) i=l
• 2p. A
given the pressure drawdown (for a well of fmite conductivity,
Note that a laminar friction factor could just as easily be used the drawdown Plat the downstream end of the well is
if the Reynolds number is less than about 2000; in the cases specified).
run under conditions applicable to high permeability
completions, it is found that the Reynolds number in the well PROBLEM IB: Minimize the (downstream) drawdown, Pi'
exceeds this threshold just a small fraction of L downstream given the production rate Q.
of the well's end. Moreover. Briggs21 recently incorporated
the effects of two-phase flow in a semi-analytic discretization PROBLEM Ie: (fmite conductivity wells only) Minimize the
of Dikken's model, and found that the pressure drop in the pressure drop inside the well,
well can be satisfactorily modeled with a single phase friction
fac1Dr-with surface roughness. as in (9).
I1P=PN -1i (14)
given the production rate Q (or drawdown P 1).
Coupling of Reservoir and Well Models
PROBLEM 2A: Make the specific flowrate into the well (the
The well-bore pressure drop equations (7) are of the functional
inflow per unit length) as uniform as possible, given the
form
drawdown, Pl'
p=F[(2l (11)
PROBLEM 2B: Make the specific flowrate into the well (the
for the pressure and flowrate vectors. This suggests a simple inflow per unit length) as uniform as possible, given the
computational scheme to solve the coupling of the reservoir production rate Q.
and the well-bore models. Equations (5) for the flowrates in
terms of the pressures at the perforations, and equations (11) All problems are subject to the constraint
in the wellbore. constitute a well-posed mathematical
problem of 7N equations in 7N unknowns. The following o~ Xt <...< Xi < X +1 <...< X N ~ L
i (15)
iterative scheme is therefore proposed: The problems of type 1 fall into the general class of problems
QII+1 =A-1P" (l2a) discussed by Inayat-Hussain19, where an optimality theorem
is proved for the PI of a reservoir-well system.
pll+l = F[Q"+1 l (12b)
Instead of doing an optimization over N variables. the above
Starting with an initial guesses P p>.p d' the application of problems can be reduced in computational size by segmenting
(12a) followed by (12b) produces an updated pressure and the perforated interval into J segments. each with I
perforation flowrate profile along the well. The method is

569
4 OPI1MlZATION OF PERFORATION DISTRIBUTION FOR HORlZONTAL WEllS SPE23005

perforations (N=J/). i.e., the perforation density is piecewise production in the upstream section of the well due to the
constant over the intervals decreased drawdown· This leads to an oveIJlrediction of the
wellrate by 27% if an infinite conductivity assumption is
used. Note that the diagram of specific inflow also shows the
increase in inflow at the ends of the well due to the increased
in which the I perforations are located at drainage volume there.
X/j+i=Xj+(Xj+l-Xj)i/I, i=l... I, i=0... J-1
In his work on analytic modeling of horizontal wells,
(17)
Dikken8 poses the question of finding the optimal well length
Note that if 1=1, then X =X·, and the optimization is over all L, given its specific productivity J S' Here optimality means
J J •
perforations independently. When 1>1, the segmentmg some fraction of the maximum flow (80% of that from a semi-
procedure reduces the computational work required by reducing infmite well) is achieved. Dikken's criterion can be evaluated
the number of equations from N to J, where for instance N may for the current perforation model, where J s has been calculated
typically be 400, but J =20 is able to deliver a reasonably as Q/(LP d) from the infmite conductivity well of Figure 1. It
smooth solution. is found that for the parameters given in Table I, the critical
well-length (when Dikken's xn=l) is approximately 900
PROBLEMS 1A and 1B are equivalent in the case of an infmite meters (for a perforation density of 1 shot per meter, and
conductivity well (as are 2A and 2B), but differ when the friction exponent a=0.2), so that the well lengths considered
pressure drop in the wellbore is significant, when the pressure here are less than the maximal and should be considered of
versus flowrate relationship becomes nonlinear. The imite length.
solution to PROBLEM 1A is of interest in the case when water
or gas coning is not a consideration. PROBLEMS 1B and 1C Accordingly, Dikken's semi-analytic imite length well model
are of interest because exaggerated coning of water or gas is can be compared to the imite conductivity perforated well
likely to occur at the downstream end of the pipe if there is a model described above. In Dikken's original model the
relatively large pressure minimum there. This phenomenon downstream drawdown is assumed known; in the course of the
has been observed in the reservoir simulations of Korady et. current work the Dikken model has been extended to allow the
al. 13, where a streamtube model shows rapid breakthrough at specification of flowrate. Figure 3 shows the drawdown along
the downstream end of a horizontal well. Also, in PROBLEM the 400 m well for both models, given a specified production
1C, a reduction in AP will lead to an increase in the rate of 10 000 bbllD [1590 m 3 /d]. The agreement is shown to
productivity index. PROBLEMS 1 are solved by a quasi-Newton be quite good, given that the two parameters in Dikken's
method with finite difference approximations for the model, J s and a, were chosen a priori , the former from the
gradient, which is a standard procedure for imding the local infmite conductivity well of the same length and perforation
minimum/maximum of a function of J variables22 • density, and the latter by attempting to match the friction
factor (9) with Dikken's power-law friction factor. It is
PROBlEMS 2A and 2B are more relevant when retardation in apparent that a closer match of the results could be achieved
coning is the prime consideration; a uniform flowrate into the by decreasing a slightly, which strongly influences the well
well is likely to delay breakthrough, attempting to make the pressure drop, and then adjusting J s to match the drawdown.
two-phase interface approach the well uniformly along its
length. The specific flowrate (flowrate per unit length) is the
Optimization of Infinite Conductivity Wells
same into each segment when
In this section, the pressure drop along the well is assumed
/~1) Q .
LQ, =-(Xj+l-Xj ), J=0... J-1 negligible compared to the drawdown, so that Pi = P d' the
(18) drawdown (e.g. in tight reservoirs, or for shorter, larger
i=/j+l L
diameter wells). With the simple model (2) and its inversion
As Q is also an unknown, (18) constitute a system of J (5), the effect of perforation distribution can be evaluated for
equations in J unknowns (Xl .. X J-l and Q). Q can be a imite length well in an isotropic infmite reservoir, without
eliminated from the system, leaving J -1 equations in as many the effects of wellbore pressure drop. The results isolate the
unknowns. Framed as such, PROBLEM 2 is not solved as an mechanisms involved in each of the two main prototype
optimization problem, but has a unique solution which is optimizations described, which will become important in the
arrived at using the iterative Newton's method. next section when wellbore pressure drop is accounted for.

Uniformly Perforated Wells Figure 4 shows the results of optimizing the flowrate
(PROBLEM 1A) and the results of making the specific inflow
Before carrying out any perforation optimizations, it is worth
constant (PROBLEM 2A) for an infmite conductivity 200 m
examining the results of the well-reservoir model with and
well, perforated with 400 perforations (for computational
without wellbore pressure drop, when the perforations are
purposes, J=10 segments were used for 1A, and J=20 for 2A).
uniformly spaced so that
Note that a 100 mD reservoir and a 100 kPa drawdown has
IXi +l -Xi l=L/(N-l), i=l...N-l (19) been used in this section, so that the well productivity is
relatively low, and the pressure drop in the wellbore is
Figure 2 compares both infinite and imite conductivity negligible. This may be shown from direct calculations,
models, for a 400 m well in a 1000 millidarcy reservoir (see where the finite conductivity well model predicts a 0.284 kPA
Table 1 for the other well and reservoir parameters). The pressure drop, which is only 0.3% of the drawdown, or by use
downstream drawdown is set at 50 kPa, and a 14 kPa pressure of an analytic scaling formula l9 .
drop is set up inside the wellbore as a result of the fluid flow.
Also in Figure 2, the graph of specific inflow (flowrate per Examining the solution with maximum Q ill"st (PROBLEM 1A),
unit meter into the well) shows there is a significant drop in the optimal solution tends to distribute more perforations at

570
SPE23(105 M. J. LANDMAN AND W. H. GOIDI'HORPE 5

the ends of the well. where there is greater exposure to the change in drawdown of the wells as production is varied, with
reservoir. Note. however. that the increase in productivity is the ratio !iP/P 1 (indicating the importance of pressure drop in
only 0.13% over that of the 1295 bbl/D [206 m 3 /d] produced the well) increasing with production rate. At high production
by a uniformly perforated well. rates. the specific inflow in the uniformly perforated well is
seen to peak very significantly at the downstream end of the
Making the specific inflow more uniform has the opposite well, which can be remedied with a significant decrease of
effect on perforation distribution. Perforations are moved perforation density there.
away from the ends of the well, reducing the enhanced
productivity there due to partial penetration effects. The The perforation density in PROBLEM 2 reaches a maximum
resulting decrease in well productivity is again small (a 0.6% approximately 75% of its length upstream from the exit, as
loss). seen in the profiles of Figure 6. In practice. it may not be
necessary to decrease the perforation density in the froal
The above results suggest that in homogeneous formations section of the well. This is because a higher specific inflow
for wells in which the pressure drop is negligible. there is may be desirable at the well ends due to the increased drainage
little advantage in non-uniform perforation distributions in volume there. Reservoir simulations coupled with a stream
order to optimize productivity. As a result, however. the tube model13 demonstrate this effect for infmite conductivity
high inflows at a well's end can be reduced by distributing less wells. where water breakthrough occurs in the center of the
perforations at the end of the well. without a significant loss completion. There is therefore scope to modify the constant
in productivity. specific inflow requirement near the well extremities in order
to aChieve uniform cresting of an underlying aquifer along the
Optimization of Finite Conductivity Wells entire well length.
The solutions to PROBLEMS lA (flowrate optimization) and 2A
Reservoir with Two Permeability Zones
(uniform inflow) for a 400 meter well with 400 perforations
in a 1000 md reservoir are shown in Figure 5. and are In order to evaluate the effect of a permeability contrast in
compared to a uniformly perforated well. Pressure drop is the formation, a two-zone permeability model is used, where a
calculated in each of the wells. Recall that the uniformly vertical plane at position x=xo separates a region M 1 of
perforated well was compared earlier to that with infmite permeability kl for x<Xo• and a region Mz of permeability "z
conductivity. and well-pressure drop was found to be for x>xo' The equations for the pressure in terms of the inflow
important in this case. for the two-zone model are given in Appendix B. They are of
the same form as those for the homogeneous formation, so
In order to maximize the inflow for PROBLEM lA, perforations that the iterative schemes described above can be used.
are distributed more densely at the downstream end of the
well. where the drawdown is largest. Note that although the Figure 7 shows the results of a 400 m well penetrating 200 m
perforation density is more than doubled at the bore outlet, of a 2000 md zone and 200 m of a 500 md zone. with the less
the improvement in production is only 2% over that of the permeable formation at the upstream end of the well. If a
uniformly perforated well. constant specific productivity well were assumed. a naive
calculation would suggest that 80% of the perforations should
In general. for all problems of type 1 (maximize Q, or be shot in the less permeable zone if the aim is to equalize the
minimize drawdown). changes in productivity are not large, inflow in each zone (a simplified version of PROBLEM 2B).
and tend to put more perforations where the drawdown is The zonal production for a set of different perforation
largest. As the optimized configuration would probably strategies is shown in Figure 7. each with 400 total
enhance coning due to significantly nonuniform specific perforations. and uniform (but generally different) perforation
inflow. the current calculations suggest that this type of densities within each zone. The production rate is kept
perforation distribution is generally unfavourable for those constant at 10 000 bbl/D [1590 m 3 /d]. The pressure drop in
reservoirs where coning is expected. the well tends to deplete the upstream inflow, so that in the
case shown 86% of the perforations should be shot in the less
The solution to PROBLEM 2A in Figure 5 distributes the permeable zone to equalize the inflow.
perforations so that their density towards the end of the well
is almost twice that of the downstream end. Note that the Similarly, PROBLEM lA (which maximizes the production) has
density drops off slightly at the upstream end due to the been solved for two-zone reservoirs. In general. a high
partial penetration effect which causes a greater inflow per density of perforations should be placed in the more
perforation at the well extremity. The decrease in wellrate is permeable zone in this case. Perforations are only necessary
only 3% as a result of redistributing perforations away from in the less permeable zone if the permeability contrast is
the exit of the well. mild, or once the perforation density becomes sufficiently
high in the more permeable zone to cause considerable
This solution for a specific-inflow-optimized finite pressure drawdown depletion there.
conductivity well is perhaps of most interest in practice, and
quantifies the field procedure of increasing perforation The two-zone model can of course be generalized to multiple
density. at the end of a horizontal well. Figure 6 shows the zones of different permeability. Similarly, boundaries and
operation of horizontal wells in a 1000 md reservoir. more complex anisOlropies in the reservoir could be included
operating under three specified flowrates of 5. 10 and 15 in a more sophisticated reservoir-well model, creating more
thousand bbl/D [795. 1590, 2380 m 3 /d]. At each flowrate a realistic forms of the pressure-flowrate relationship (4).
well with uniform perforation density and a well with a
perforation profile optimized for uniform specific inflow
(PROBLEM 2B) are shown. The plot of pressure indicates the

571
6 OPl1MlZATION OF PERFORATION DISTRIBUITON FOR HORlZONTAL WELlS SPE23005

Conclusions matrix element


Dikken's friction exponent
• A model has been presented that allows the optimal
perforation radius (m)
perforation density for a horizontal well producing from an
wellbore rouglmess (m)
idealized infinite reservoir to be quantified. In a real
dynamic viscosity (Pa s)
homogeneous reservoir, the results should be qualitatively
density rr-g/m3 )
the same.

• For a finite conductivity well, making the specific inflow


uniform along the well requires a perforation density which
AcknOWledgments
increases towards the upstream end of the well. Alternatively,
if maximum production is sought, the perforation density at Dr. A. A. Inayat-Hussain and Dr. B. P. Marett are thanked for
the downstream end of the well should be increased. The total many helpful discussions during the course of this work. BHP
flowrate remains relatively insensitive to perforation is acknowledged for permission to publish this paper.
distribution. however.
References
• The results of Dikken! have been reproduced for a uniformly
1. Goode, P.A. and Thambynayagam, R.K.M.: "Pressure
perforated finite conductivity well.
Drawdown and Buildup Analysis of Horizontal Wells in
Anisotropic Media," SPE Reservoir Engineering, Dec. 1987,
• A simple application of the model to a well penetrating two
683-697.
zones of contrasting permeability has been given. The model
can be enhanced to determine optimum perforation 2. Kuchuk, F.I., Goode, P.A., Sherrard, D.W., Brice, B.W.,
distributions within multiple zones, and when reservoir and Thambynayagam. R.K.M.: "Pressure Transient Analysis
boundaries (no-flow or constant pressure) are included. and Inflow Performance for Horizontal Wells," SPE 18300
presented at the SPE Annual Technical Conference and
• More sophisticated perforation inflow models could be added Exhibition. Houston, Texas, Oct. 2-5, 1988.
in order to allow for damaged zones, the presence of the 3. Kuchuk, F.I., Goode, P.A., Wilkinson, 0.1. and
wellbore 23 , and permeability anisotropy. Similarly, a
Thambynayagam, R.K.M.: "Pressure Transient Behavior of
multiphase wellbore pressure drop model could be included. Horizontal Wells with and without Gas Cap or Aquifer," SPE
Reservoir Engineering, March. 1991, 86-94.
• As the model allows for the determination of the pressure
field throughout the reservoir, there is potential to directly 4. Ioshi, S.D.: "Augmentation of Well Productivity with
calculate the effect of perforation distribution on fluid flow in Slant and Horizontal Wells," JPT (June 1988) 729-739.
the formation. 5. Goode, P.A. and Willcinson. 0.1.: "Inflow Performance of
Partially Open Horizontal Wells," SPE 19341 presented at the
SPE Eastern Regional Meeting, Morgantown. West Virginia,
Oct. 24-27, 1989.
Nomenclature
6. Giger, F.M.: "Analytic 2-D Models of Water Cresting
a perforation length (m) before Breakthrough for Horizontal Wells," SPE Reservoir
A cross-sectional area of wellbore (m2 ) Engineering, Nov. 1989, 409-416.
A matrix relating P to Q
7. Babu, O.K. and Odeh, A.S.: "Productivity of a Horizontal
D wellbore diameter (m)
Well," SPE Reservoir Engineering, Nov. 1989,417-421.
1,Ii single phase friction factor
F pressure drop functional 8. Dikken, B. I.: "Pressure Drop in Horizontal Wells and its
G flowrate functional Effect on Production Performance," JPT (Nov. 1990) 1426-
I number of perforations in a pipe segment 1433.
J number of uniformly perforated segments
9. Landsberg, 1.1. and Fowkes, N.D.: "Water Movement
k permeability (m2 )
Through Plant Roots," Ann. Bot. 42 (1978), 493-508.
/q permeability in zone i (m2 )
Js specific productivity index (m3 /Pa/m) 10. Stone, T.W., Edmunds, N.R. and Kristoff, B.I.: "A
L length of perforated horizontal well (m) Comprehensive Wellbore/Reservoir Simulator," SPE 18419
qi flowrate in wellbore between perf. i and i+l (m3 /s) presented at the SPE Symposium on Reservoir Simulation,
Q total flowrate from well (m3 /s) Houston, Texas, Feb. 6-8, 1989.
(Jj flowrate into ith perforation (m3 /s) 11. Islam, M.R. and Chakma, A.: "Comprehensive Physical
Q vector of perforation flowrates (m3 /s) and Numerical Modeling of a Horizontal Well," SPE 20627
N total number of perforations presented at 65th Annual Technical Conference of the SPE,
Pi pressure at ith perforation (Fa) New Orleans, Louisiana, Sept. 23-26, 1990.
P..tl downstream drawdown (Fa)
P vector of perforation pressures (Pa) 12. Collins, D.A., Ngiem, L.X., Sharma, K.A. and Kamal,
Rei local Reynolds number N.I. "Field Scale Simulation of Horizontal Wells with Hybrid
V arbitrary vector of length N Grids," SPE 21218 presented at the 11th SPE Symposium on
Xi position of ith perforation (m) Reservoir Simulation, Anaheim, California, Feb. 17-20,
Xi position of ith perforation segment (m) 1991.

572
SPE23005 M. J. LANDMAN AND W. H. GOLDTHORPE 7

13. Korady, V., Renard, G. and Lemonnier, P.: "Modelling of Appendix B:


Pressure Drop for Three-Phase Flow in Horizontal Wells," Two Zone Reservoir
presented at the 6th European lOR-Symposium, Stavanger,
A vertical plane at position x=xo separates a region M 1 of
Norway, May 21-23, 1991.
permeability 11 for x<xo, and region M 2 of permeability Ie"
14. lchara, M.J.: "The Performance of Perforated for x>xo' The well is assumed to penetrate both zones. The
Completions in Gas Reservoirs," SPE Reservoir Engineering, equation for the pressure at perforation i in zone M 1 is
Nov. 1990, 639-646. calculated by imaging each perforation across the interface.
The image of the perforation itself is also taken as a point
15. Bell, W.T., Brieger, E.F. and Harrigan, J.W.: "Laboratory
sink of appropriate strength, as the perforations are assumed
Flow Characteristics of Gun Perforations," iPT (Sept. 1972)
to be sufficiently far from the permeability interface. The
1095-1103.
total pressure at such a perforation is therefore
16. Muskat, M.: 'The Effect of Casing Perforations on Well
Productivity," Trans. AIME (1943) 151, 175-187.
17. Besson, 1.: "Performance of Slanted and Horizontal Wells , =...1!:....
p(l) 4nk [aJl... ~ + ~
"'" a~!)Q.
'J J
+ "'"
~
a~~)Q']
'/ J
1 jeMI jeM2
on an Anisotropic Medium," SPE 20965 presented at Europec
90, the Hague, Netherlands, Oct. 22-24, 1990.
(Bl)
18. Lee, S.H.: "Analysis of Productivity of Inclined Wells
and its Implication for Finite-Difference Reservoir
Simulation," SPE Production Engineering, May. 1989, 173-
180.
19. Inayat-Hussain, A.A.: "The Productivity Index of
Petroleum Reservoirs and the Dissipation Function," BHP
Research preprint (1991).
20. Miller, D.S.: Internal Flow Systems, BHRA (Information
Services), Cranfield (1990)
21. Briggs, P.: "Pressure Drop in Horizontal Well
Completions - Is it Important 1" 1991 JETEC Conference.
22. Press, W.H., Flannery, B.P., Teukolsky, S.A. and A similar expression holds for the pressure at a perforation in
Vetterling, W.T.: Numerical Recipes, Cambridge Uni. Press, zone 2, and the problem is therefore of the form (4), as for the
Cambridge (1989) homogeneous reservoir.
23. Marett, B.P., Inayat-Hussain, A.A. and Goldthorpe,
W.H.: "Analytical Solutions of 3-Dimensional Flow near a
Wellbore," SPE 23008 to be presented at the SPE Asia Pacific
Conference, Perth, Australia. Nov. 4-7, 1991.

AppendiX A:
Iterative Scheme When Flowrate Is Specified
When Q is specified, the iterative scheme is altered so that
(12a) is replaced by

Q"+1 = G[A-11'''] (AI)


where

(A2)

is a scaling function for its vector argument. This ensures


that the flowrate is correct at each iteration. In the second
step (12b), equation (7a) is no longer available, but the set
(7b) is augmented with equation (2), which is evaluated at a
perforation midway along the wellbore (this is more stable
than picking a perforation at the end of the bore). For an
inImite conductivity well, (AI) is applied to a constant
pressure vector, and due to the linearity of the problem, the
resulting perforation flowrates and the pressure vector can be
scaled to give the correct total flowrate.

573
SPE 2 3005
Table 1. Reservoir Fluid and Well Parameters

ReservoIr fluid parameters Well parameters

viscosity 0.5 centipoise well diameter D 12.5 cm


J.L
density p 750 kg/m3 well roughness 1C 0.1 cm
perf. radius 0 1.0 cm
perf. length a 35.0cm

X=Xi x=xi+l X=X N

Fig. 1 - Perforated wellbore model

........ 30 ~_ _---I. L-_ _..l...-_ _-L


-35
.g
'2 -40
g 25
;g .. - -- -_ - -_ _.. ---_ -_ .. '
~ uniformly perforated wells:
- - - - infinite conductivity
'-' 20
~
~ -45 - fmite conductivity
S
'"'"~ (k= 1000md) .S 15
c..>
t=l.< -50 ---- infinite conductivity (Q=8775 bbl/D)
!E 10
- fmite conductivity (Q=6930 bbl/D)
~
-55 ........----.-----,---~--.J_ ~5-;----,----r----.----f-
o 100 200 300 400 o 100 200 300 400
x(m) X (m)

Fig. 2 - Comparison of infinite and finite conductivity


uniformly perforated wells

-50

-55
'2
~
-60

~ -65
'"
a. -70 .---- Dikken model, a.=0.2, J=0.439 bbl/D/kPa/m
- current model, 400 uniform perforations
-75

-80
0 100 200 300 400
x(m)

Fig. 3 - Comparison of Dikken's and current model

574
seE 2"3005

8 2 .6
-I
..-..
.g
11
~
&2.4 I g 10 .......... unifonn perforations
I

,
I
'-"
>- 2.2
I
I ;g 9
. - - flowrate optimized I
.<;:::; '-" - unifonn specific inflow
L , l
=
r:IJ
~ , ~
0 8
II
IJ
~ 2.0 l:i=:
-------- Ii
=0
'p 1.8
.S
(.)
7 I,,;i
ca
.... $(.) . .:::..
-:;,.""":.:~-..:.-::-.:.-~ ,.=-:::~
6
.g 1.6
(I)
(I)
p.,
r:IJ
p., 5
0 50 100 150 200 0 50 100 150 200
X (m) x (m)

Fig. 4 - Optimization problems 1A and 2A for


an infinite conductivity well

..-.. ..-.. 50 -.r---'-----'-------'---ooo\-


:€. 2.0 - I
.g
'gp., : g40~
'>: 1.5 .. - ;g \..
•<;:::; '-" 30 I
r:IJ
=
~ 1.0
::>
;>
r.:8
'.
~-
. .,
I
- ""
§ '-------_ . .~ 20 ~~:::.i:.:.-.::...;..-::-.::.::.::_::.:T:.:...,....I
.~ 0.5
.... $ 10
.g
& 0.0 -.,------,.------r-----,----r ~ O-.,------,.------r-----r----r
o 100 200 300 400 o 100 200 300 400
x(m) X (m)

-36 _- _--_._ .

----
"2
-38 ~~.
...
---------
~ uniform perforations 6930 bbl/D
~
-40
-42 flowrate-optimized 7090 bbl/D
~
r:IJ -44 uniform specific inflow 6720 bbl/D
r:IJ
e
p., -46
k= lOOOmd
-48
-50
0 100 200 300 400
x (m)

Fig. 5 - Optimization problems 1A and 2A for


a finite conductivity well

575
SPE 2 '3005

e 1.4 ".......
8 70
'i3" -.....;.
goo - uniform perforations
._--' uniform specific inflow
8.1.2
......,
~
'r;: 1.0
e ~ 40
SO

5 o
"'t::) 0.8
........ uniform perforations
uniform specific inflow: E30 --...-... _- .... _---------- ---"..
... ...
....~6 0.6 - - . 5 000 bbl/D
$
(,) 20
·····10000 bbl/D ... _~---- ... -
~8. 0.4 -\------..,.-----,r-----,----r-
- 15000 bbl/D
8. 10O-\-----..,.-----,r-----,----r-
UJ
5000bbl/D

o 100 200 300 400 o 100 200 300 400


x(m) x(m)
-20 J-_ _-l.._ _.....L.._ _.......L._ ____\__

-40 5000bbl/D
--==----~
-----------------

- uniform perforations
-140 .•••. uniform specific inflow k = 1000 md'
o 100 200 300 400
x(m)

Fig. 6 - Optimization of specific inflow (problem 2B)


at three wellrates

100

........d
0
(,)
80
.g
0
~ 60
=CI,)
~ ~""""'--+I--'"
40
....
';i
:L
....0
'+-t
.x=2
0 20
~

0
0 20 40 60 80 100
% of perforations in zone 2

Fig. 7 - Production of a well penetrating two zones,


for different perforation distributions

576

You might also like