Professional Documents
Culture Documents
Generalized Horizontal Well Inflow Relationships for Liquid, Gas, or Two-Phase Flow
R. Kamkom and D. Zhu, Texas A&M U.
7.08 × 10 −3 k H h( p e − p wf )
q= (
7.08 × 10 −3 kL p e − p wf )
⎛ ⎛ ⎞ I h ⎛ I h ⎞ ⎞ qo =
⎜ a + a − (L / 2 )
2 2
⎟
μ o Bo ⎜ ln⎜⎜ ⎟ ⎜ ⎟
⎟⎟ + L ln⎜ r (I + 1) ⎟ + s ⎟
ani ani
⎛ ⎡ hI ani ⎤ πy b ⎞
⎜ ⎜ (L / 2)2 ⎝ w ani ⎠ ⎟ μ o Bo ⎜⎜ ln ⎢ ⎥+ − 1.224 + s ⎟
⎟
⎝ ⎝ ⎠ ⎠ ⎝ ⎣ rw ( I ani + 1) ⎦ hI ani ⎠
(1)
(5)
where,
where, k is defined as k = k H kV , and Eq. 5 can be
rearranged
kH
I ani = (2)
kV
(
7.08 × 10 −3 k H L p e − p wf )
and qo =
⎛ ⎡ hI ani ⎤ πy b ⎞
μ o Bo ⎜⎜ I ani ln ⎢ ⎥+ − I ani (1.224 − s ) ⎟
⎟
⎧ 0.5 ⎫
0.5 ⎝ ⎣ rw ( I ani + 1) ⎦ h ⎠
L⎪ ⎡ ⎛ reH ⎞ ⎤ ⎪
4
⎢
a = ⎨0.5 + 0.25 + ⎜⎜ ⎟⎟ ⎥ ⎬ (3) (6)
2⎪ ⎢ ⎝ L / 2 ⎠ ⎥⎦ ⎪
⎩ ⎣ ⎭ Notice that with different approach, Butler’s model and
Furui’s model yield very similar expression for inflow
performance for horizontal wells. If kx is close to ky, at skin
In Eqs. 1-3, kH is horizontal permeability, and kv is vertical factor equal to zero, the only slight difference is the constants
permeability. The geometry parameters are defined in Fig.1, in the denominators (1.14 for Butler and 1.224 for Furui).
and the other variables are defined in the nomenclature. When anisotropic ratio increases, the results from the two
Whenever Iani is used in this paper, it is assumed that kx and ky models deviate slightly. Fig. 3 shows the plots of IPR curves
are equal and replaced by kH. generated by Butler’s and Furui’s models at an Iani of 3
without skin effect.
More-recent models are developed with different
approaches in drainage geometry. Butler3 presented an IPR
model for horizontal wells under the steady-state condition. Compressible Fluid (Gas Well). For horizontal gas well
The model predicts the productivity of fully penetrated deliverability, the model can be derived similar to the one for
horizontal well in a box-shaped reservoir. This model can oil wells. The modifications needed are the gas formation
handle either isotropic or anisotropic reservoir. Fig. 2 shows volume factor which is a function of pressure and temperature,
the geometry used to develop the equation. The model was and the unit of flow rate. Gas wells usually have high velocity
obtained by applying the superposition principle. Butler’s so that non-Darcy flow effect should be considered. We
model can be written as, modified Furui’s model for gas wells. First, the unit of IPR
equation for oil wells needs to be converted for gas wells
(from STB/day to Mscf/day)
(
7.08 × 10 −3 k H L p e − p wf )
qo = (4)
⎛ ⎡ hI ani ⎤ πy b ⎞ 1 ⎛ STB ⎞ 1000 ⎛ Mscf ⎞ ⎛ Mscf ⎞
μ o Bo ⎜⎜ I ani ln ⎢ ⎥+ − 1.14 I ani ⎟
⎟ q o ⎜⎜ ⎟⎟ ⎜ ⎟ = 25146.928q g ⎜⎜ ⎟⎟
⎝ ⎣ rw ( I ani + 1) ⎦ h ⎠ 7.08 × 10 −3 ⎝ day ⎠ 5 . 615 ⎝ STB ⎠ ⎝ day ⎠
(7)
4 Then we need to relate the gas formation volume factor, Bg, to
Furui, et. al. also presented an analytical model for fully
penetrated horizontal well in a box-shaped reservoir using the pressure and temperature. By the Real Gas Law,
same system shown in Fig. 2. It assumed that the flow to a
horizontal well can be divided into two regimes, a radial flow
region near the wellbore and a linear flow region away from znRT / p
Bg = (8)
the wellbore. On the top and bottom of the reservoir, they used z sc nRTsc / p sc
no flow boundary condition. The model can be used to
calculate isotropic and anisotropic reservoir. The well is
located in the center of the reservoir. A skin factor was added where Tsc and psc are the pressure and temperature at the
to the model to include the effect of formation damage on well standard condition. If the pressure and temperature at the
productivity. This model was developed based on the standard condition are 14.7 psi and 520ºR respectively, and
simulation results of a finite element model (FEM) for the gas compressibility factor, Z, at the standard condition is 1,
incompressible fluid. The IPR equation of Furui’s model is we can rewrite Eq. 8 as
SPE 99712 3
(10)
where, the turbulent factor for undamaged and damaged
Substituting Eq. 10 into Eq. 5 and changing oil viscosity to gas zones, β and βd, are estimated by6
viscosity at an average pressure, the IPR equation for
horizontal gas wells can be expressed as
2.6 × 1010
β= (16)
(
kL p e2 − p wf
2
) ( kxkz )1.2
qg =
⎛ ⎡ hI ani ⎞ and
⎤ πy b
1424 z μ g T ⎜ ln ⎢ + − 1.224 + s ⎟
⎜ r ( I + 1) ⎥ hI ⎟ 2.6 × 1010
⎝ ⎣ w ani ⎦ ⎠ βd =
( )d 1.2
ani (17)
(11) kxkz
qg =
( ( ) ( ))
kL m p − m p wf
(13)
recovery factor and a reservoir pressure of 4023 psi, the
analytical results match the simulation results when the
⎛ ⎡ hI ani ⎤ πy b ⎞ pressure drawdown is below 1200 psi.
1424T ⎜ ln ⎢ ⎥+ − 1.224 + s ⎟
⎜ r (I ⎟
⎝ ⎣ w ani + 1) ⎦ hI ani ⎠
Pseudo-Steady-State Condition. Pseudo-steady-state
condition assumes that there is no flow at the boundary of a
For gas wells, the flow velocity is usually much higher reservoir domain. An average reservoir pressure, p , is
than the one for oil wells, especially near the wellbore. This introduced in the IPR equations. Pressure decline curves, if
high velocity causes additional pressure drop, which is known available, can be used to calculate the average pressure as a
as non-Darcy flow effect. This additional pressure drop is a function of time, and therefore to obtain a production history.
function of flow rate, and can be added to Eq. 13 as
qg =
( ( ) ( ))
kL m p − m p wf horizontal well IPR model under the pseudo-steady-state
condition. The model rotates a vertical well to represent a
⎛ ⎡ hI ani ⎤ πy b ⎞ horizontal well, and uses a shape factor to count for the
1424T ⎜ ln ⎢ + − 1.224 + s + Dq g ⎟
⎜ r ( I + 1) ⎥ hI ⎟ drainage area change and a partial penetration skin factor for
⎝ ⎣ w ani ⎦ ani ⎠ partial penetrated wellbores. Fig. 5 shows the geometry model
(14)
4 SPE 99712
used by Babu and Odeh. The model can handle both isotropic k y k z ⎡⎛ 1 x ⎞ ⎤
6.28b 2 x2 ⎟ + L ⎛⎜ L − 3 ⎞⎟⎥
and anisotropic reservoirs and the well can be in any positions Py = ⎢⎜ − mid + mid
in a box-shape reservoir. Babu and Odeh’s IPR equation for ah k x ⎢⎜⎝ 3 b b2 ⎟ 24b ⎝ b
⎠ ⎠⎥⎦
⎣
horizontal well at the pseudo-steady-state condition is
(24)
and
(
7.08 × 10 −3 b k y k z p − p wf )
qo = (18) ⎛b ⎞⎛ 6.28a ⎞⎛ 1 y y2 ⎞
⎡ ⎛ A 0.5 ⎞ ⎤ Pxy = ⎜ − 1⎟⎜ k z k y ⎟⎜ − 0 + 0 ⎟ (25)
μ o Bo ⎢ln⎜⎜ ⎟ + ln C H − 0.75 + s R ⎥
⎟ ⎝ L ⎠⎝ h ⎠⎜⎝ 3 a a 2 ⎟⎠
⎣⎢ ⎝ rw ⎠ ⎦⎥
ln C H = 6.28
a ⎡1 y ⎛y ⎞ ⎤
2
⎢ − 0 + ⎜⎜ 0 ⎟⎟ ⎥
⎧
[ ]
⎪− (x ) 0.145 + ln (x ) − 0.137(x ) for x = 2b , x =
2 L 4 x mid ± L
2b
≤1
I ani h ⎢ 3 a ⎝ a ⎠ ⎥ ⎪
⎣ ⎦ (19) ⎨
⎛
− ln⎜⎜ sin
πz 0 ⎞ ⎡ a ⎤
⎟⎟ − 0.5 ln ⎢ ⎥ − 1.088
⎪
⎩
[ ] 4x
⎪(2 − x ) 0.145 + ln (2 − x ) − 0.137(2 − x )2 for x = mid
2b
±L
>1
⎝ h ⎠ ⎣ I ani h ⎦ (26)
where
For fully penetrated wellbore (the wellbore length L is equal
to the drainage length b in Fig. 5), the partial penetration skin
is zero. For partially penetrated wellbore (wellbore length L is x mid = 0.5(x1 + x 2 ) (27)
shorter then the drainage length b in Fig. 5), the equation used
to calculate sR depends on the geometry and the permeability
anisotropy. If a k y ≥ 0.75b k x >> 0.75h k z , we have As pointed out in the original paper, the model assumed that
the thickness of the formation, h, is generally much smaller
than the other two dimensions of the drainage box, a and b. If
' this condition does not apply, it should be examined first if a
s R = Pxyz + Pxy (20) horizontal well is the right application for the field. The
equation is very helpful when used to examine the effects of
the reservoir and well parameters on well performance, and
and if b k x > 1.33a k y >> h k z , then sR is therefore to optimize well design and operation. Fig. 6 shows
the IPR curves generated by the Babu and Odeh’s model for
different anisotropic ratio. From the plot, we can see that at
high anisotropic (high difference between vertical and
s R = Pxyz + Py + Pxy (21) horizontal permeability) the production rate is low comparing
with isotropic reservoir. The flow rate reduces rapidly when
the vertical permeability is less than 40 percent of the
In Eqs. 20 and 21 horizontal permeability. Even all developed for single-phase
oil wells, the results from the Butler model and the Furui
model should not be compared with the Babu and Odeh model
⎛b ⎞⎡ h ky ⎛ πz ⎞ ⎤ since the assumptions of the boundary conditions used to
Pxyz = ⎜ − 1⎟ ⎢ln + 0.25 ln − ln⎜ sin ⎟ − 1.84⎥ (22) develop the models are different.
⎝ L ⎠ ⎢⎣ rw kz ⎝ h⎠ ⎥⎦
(23)
SPE 99712 5
b k y k z ⎛⎜ p − p wf2 ⎞
2 Examples and Discussions
⎟
⎝ ⎠
qg = (28)
⎡ ⎛A ⎞ 0.5 ⎤ Oil Well. A 4-inche diameter horizontal well fully penetrates
1424 z μ g T ⎢ln⎜ ⎟ + ln C H − 0.75 + s R ⎥
⎜
⎢⎣ ⎝ rw ⎠ ⎟ in the length direction of the box-shaped oil reservoir. The
⎦⎥
reservoir size is 2000 ft in length, 1500 ft in width and 100 ft
in height. The horizontal permeability is 100 md and the
The gas properties are estimated at the average pressure vertical permeability is 10 md. The skin factor is zero. The oil
between the flowing bottomhole pressure and the reservoir viscosity is 2 cp and the oil formation volume factor is 1
pressure. Either squared pressure or the real gas pseudo- resbbl/STB.
pressure function can be used to calculate the gas well For steady state, we can use either the Butler model or the
deliverability for horizontal wells. To use the real gas pseudo- Furui model to predict the well performance. Furui’s model is
pressure, and considering non-Darcy flow effect, the equation used in this example. We first calculate the anisotropic ratio,
becomes Iani, and the average permeability for Furui’s model
( ( ) ( ))
b k y k z m p − m p wf
I ani =
kH
=
100
= 3.16228
qg =
⎡ ⎛ A 0.5 ⎞ ⎤ kV 10
1424T ⎢ln⎜ (
⎟ + ln C H − 0.75 + s R + b s + Dq g )⎥
⎢⎣ ⎜⎝ rw ⎟
⎠ L ⎥⎦ k = k H kV = (100)(10) = 31.6228
(29)
Then the productivity index, qo/(pe-pwf), is
Two-Phase Flow Well. For two-phase, the IPR of horizontal
well can be predicted by using the correlation since the
analytical model for two-phase flow is very difficult to derive qo 7.08 × 10 −3 kL
=
because of the complexity of the relative permeability. Hence, ( pe − p wf) ⎛ ⎡ hI ani ⎤ πyb ⎞
the correlations to estimate the two-phase IPR have been used
μ o Bo ⎜⎜ ln ⎢ ⎥+ − 1.224 + s ⎟
⎟
instead. Vogel's correlation9 is a popular two-phase IPR model
⎝ ⎣ rw ( I ani + 1) ⎦ hI ani ⎠
for vertical wells, and is adapted to horizontal well by using a
horizontal well single-phase oil equation to calculate the 7.08 × 10 −3 (31.6228)(2000)
maximum flow potential. The modified Vogel’s correlation is =
expressed as10 ⎛ ⎡ (100)(3.16228) ⎤ π (750) ⎞
[2][1]⎜⎜ ln ⎢ ⎥ + − 1.224 ⎟⎟
⎝ ⎣ (4 / 24 )((3.16228) + 1) ⎦ (100)(3.16228) ⎠
2
qo ⎛ p wf ⎞ ⎛p ⎞
= 1 − 0.2⎜⎜ ⎟ − 0.8⎜ wf ⎟
⎟ ⎜ p ⎟ (30) and that gives
q o,max ⎝ p ⎠ ⎝ ⎠
qo STB Day
qo,max is the maximum open flow potential, ( pe − pwf ) = 18.14 psi
qo ⎡ (150)(3.162) ⎤ π (1000)
=
(p − p ) wf
ln ⎢ ⎥ +
⎣ (0.1667 )(3.162 + 1) ⎦ (150)(3.162)
− 1.224 = 11.93
( kxkz )
1.2
( (1)(0.1) ) 1.2 ⎝ ⎠
Thus,
(2000) (1)(0.1)(m( p ) − m( pwf ))
Since there is no damage, we can calculate non-Darcy by qg =
⎡
1424(670)⎢7.75 +
⎣
2000
2000
((
⎤
8.97 × 10 −6 q g ⎥
⎦
) )
Lγ g k x k z ⎡⎛ β ⎞⎛ 1 1 ⎞⎤
D = 2.22 × 10 −15 ⎢⎜ ⎟⎜ − ⎟⎥
( )
μ g p wf ⎣⎢⎝ L2 ⎠⎜⎝ rw re ⎟⎠⎦⎥
Or in the term of pseudo pressure function,
D = 2.22 × 10 −15
(2000)(0.7 )(0.316) Two-phase Inflow Calculation. This example shows how to
(0.017 ) use the correlation presented before to calculate flow rate for a
two-phase flow well. For pseudo-steady-state condition, we
× ⎢⎜
(
⎡⎛ 1.035 × 1011 )⎞⎟⎛⎜ 1
−
1 ⎞⎤ use the productivity index from Babu and Odeh’s equation to
⎜ ⎟⎝ 0.1667 547.72 ⎟⎠⎥⎥
⎣⎢⎝ (2000 )
2
⎠ calculate the maximum open flow potential, qo,max, first. From
⎦
the single-phase, pseudo-steady-state example, we have
= 8.97 × 10 −6
qo STB Day
J= = 29.39
and
( )
p e − p wf psi
SPE 99712 7
q o,max =
(29.39)(4500) = 73475 STB ky = y direction permeability, md
1.8 day kz = z direction permeability, md
L = horizontal well length, ft
With Vogel’s correlation, the well flow rate is related to the J = Productivity for toe and heel segment, STB/psi
bottomhole flowing pressure as
m( p ) = average reservoir pseudopressure, psi2/cp
m(pwf) = flowing bottomhole pseudopressure, psi2/cp
2
qo ⎛ p wf ⎞ ⎛ p ⎞
= 1 − 0.2⎜⎜ ⎟ − 0.8⎜ wf ⎟
⎟ ⎜ 4500 ⎟
p = average reservoir pressure, psi
73475 ⎝ 4500 ⎠ ⎝ ⎠
qo = production rate, STB/d
qo,max = absolute open flow potential, STB/d
From the above equations, we can generate the IPR curve for rw = well radius, ft
two-phase flow in the horizontal well. Notice that since the
Vogel correlation is generated under pseudo-steady-state s = skin factor
condition, only the Babu and Odeh equations are used to sR = partial penetration skin at constant flow rate
calculate qo,max.
T = Temperature, ºR
x0 = position of well in x-direction
Conclusion
y1 = position of well at the heel in y-direction
This paper summarizes the inflow performance relationship
equations for gas, oil and two-phase flow in horizontal wells. y2 = position of well at the toe in y-direction
These analytical models are developed based on two boundary z0 = position of well in z-direction
conditions, the steady state condition and the pseudo-steady-
state condition. For steady-state flow, the oil well models were β = turbulence factor
developed with different assumptions about the drainage area βd = turbulence factor for damaged zone
and drainage pattern, and the gas well models were modified
γg = gas gravity
from the oil well models considering the gas properties such
as the formation volume factor of gas being a function of μ = viscosity, cp
pressure and temperature. For pseudo-steady-state condition,
the oil well model by Babu and Odeh model is more flexible
in well location in the drainage area, and it relaxed the References
assumption of fully penetrated wellbore. The model is also 1. Joshi, S.D.: “Augmentation of Well Productivity with Slant and
modified for gas wells. Two-phase correlation for horizontal Horizontal Wells,” JPT, (June 1988), 729-739.
well is also presented for the pseudo-steady-state conditions. 2. Economides, M.J., Deimbacher, F.X., Brand, C.W., and
The paper discussed the conditions each IPR equation applies Heinemann, Z.E.: “Comprehensive Simulation of Horizontal
carefully, and illustrated the procedures of using each model. Well Performance,” SPEFE, (December 1991), 418-426.
Analytical models are simpler, faster and more practical 3. Butler, R.M.: Horizontal Wells For the Recovery of Oil, Gas and
compared with numerical reservoir simulation models. When Bitumen, Petroleum Monograph No. 2, Petroleum Society of
used correctly, the models give reasonable predictions of CIM (1994).
horizontal well performances compared the results from 4. Furui, K., Zhu, D. and Hill, A.D.: “A Rigorous Formation
simulation models. The analytical models can be helpful in Damage Skin Factor and Reservoir Inflow Model for a
designing, operating and optimizing horizontal wells. Horizontal Well,” SPEPF, (August 2003), 151-157.
5. Al-Hussainy, R.R. et al.: “The Flow of Real Gases Through
Nomenclature Porous Media,” JPT (May 1966) 624-636.
A = reservoir drainage area, ft2 6. Thomas, L.K. et al.: “Horizontal Well IPR Calculation,” paper
SPE 36753 presented at the 1996 SPE Annual Technical
a = extension of drainage volume of horizontal well in x Conference and Exhibition, Denver, CO, October 6-9.
direction, ft 7. Babu, D.K. and Odeh A.S.: “Productivity of a Horizontal well,”
SPE Reservoir Engineering (Nov. 1989) 417-421.
Bo = oil formation volume factor, res bbl/STB
8. Billiter, T., Lee, J. and Chase, R.: “Dimensionless Inflow-
b = reservoir length in y-direction, ft Performance-Relationship curve for Unfractured Horizontal Gas
D = non-Darcy flow coefficient, D/Mscf Wells,” paper SPE 72361 presented at the SPE Eastern Regional
Meeting, Canton, Ohio, October 17-19, 2001.
8 SPE 99712
9. Vogel, J. V.: “Inflow Performance Relationships for Solution- 10. Kamkom, R. and Zhu, D.: “Evaluation of Two-Phase IPR
Gas Drive Wells,” Journal of Petroleum Technology, (January Correlations for Horizontal Wells,” paper SPE 93986 presented
1968), 83-92. at the 2005 SPE Production and Operations Symposium,
Oklahoma City, Oklahoma, April 17-18.
7.08 x10 −3 k H h( p e − p wf )
qo =
⎛ ⎛ ⎞ I h ⎛ I h ⎞ ⎞
⎜ ⎜ a + a − (L / 2 )
2 2
Oil ⎟ ⎟
μ o Bo ⎜ ln⎜ ⎟⎟ + L ln⎜⎜ r (I + 1) ⎟⎟ + s ⎟
ani ani
⎜ ⎜
⎝ ⎝
(L / 2 ) 2
⎠ ⎝ w ani ⎠ ⎟
⎠
Joshi-Economides (
7.08 × 10 −3 b k y k z p − p wf )
qo =
⎡ ⎛ A 0.5 ⎞ ⎤
qo =
(
7.08 x10 −3 k H L pe − p wf ) Bo μ o ⎢ln⎜
⎢⎣ ⎜⎝ rw
⎟ + ln C H − 0.75 + s R ⎥
⎟ ⎥⎦
⎠
⎛ ⎡ hI ani ⎤ πyb ⎞
μ o Bo ⎜⎜ I ani ln ⎢ ⎥+ − 1.14 I ani ⎟
⎟
⎝ ⎣ rw ( I ani + 1) ⎦ h ⎠
Butler
qo =
(
7.08 x10 −3 kL p e − p wf )
⎛ ⎡ hI ani ⎤ πy b ⎞
μ o Bo ⎜⎜ ln ⎢ ⎥+ − 1.224 + s ⎟
⎟
⎝ ⎣ rw ( I ani + 1) ⎦ hI ani ⎠
Furui
qg =
( ( ) ( ))
kL m p − m p wf ( ( ) ( ))
b k y k z m p − m p wf
Gas qg =
⎡ ⎛ A 0 .5 ⎞ ⎤
⎛ ⎡ hI ani
1424T ⎜ ln ⎢
⎤ πy b
+
⎜ r ( I + 1) ⎥ hI
⎞
− 1.224 + s + Dq g ⎟
⎟
1424T ⎢ln⎜⎜ ⎟
(
⎟ + ln C H − 0.75 + s R + b s + Dq g ⎥ )
⎝ ⎣ w ani ⎦ ani ⎠ ⎣⎢ ⎝ rw ⎠ L ⎥⎦
2
qo pwf ⎛ pwf ⎞
= 1 − 0.2 − 0.8⎜⎜ ⎟
⎟
qo,max p ⎝ p ⎠
2-φ
qo, max =
1 ()
7.08 × 10−3 b k y k z p
1.8 ⎡ ⎛ A0.5 ⎞ ⎤
Bo μo ⎢ln⎜ ⎟ + ln CH − 0.75 + sR ⎥
⎢⎣ ⎜⎝ rw ⎟⎠ ⎥⎦
SPE 99712 9
2.5E+07
2.0E+07
(pr -pwf )
2
Economides, et. Al. model
2
1.0E+07
Simulation, 10% RF
Analytical, 10% RF
5.0E+06 Simulation, 20% RF
Analytical, 20% RF
z
0.0E+00
y
0.0E+00 5.0E+04 1.0E+05 1.5E+05
L Gas Flow Rate (Mscf/day)
z Streamline
L
b
yt
yb
Fig. 5 Geometry model for pseudo-steady-state flow equation.
10000
5000
8000
4500
Flow Rate (STB/day)
2500
2000
2000
1500 0
1000 0.0 0.2 0.4 0.6 0.8 1.0
500 The Ratio of Vertical to Horizontal Permeability
0 (kV/kH)
0 20000 40000 60000 80000
Flow Rate (STB/day)
Fig. 6 Flow rate as a function of anisotropic ratio for pseudo-
steady-state equation
Fig. 3 IPR curves for steady-state flow at s=0.