Professional Documents
Culture Documents
Chapter 4 (WTA) - Hydraulically Fractured Wells (D. Tiab)
Chapter 4 (WTA) - Hydraulically Fractured Wells (D. Tiab)
ANALYSIS
Ch. 4 – Hydraulically
Fractured Wells
Djebbar TIAB, Ph.D.
Professor Emeritus, Petroleum Engineering
University of Oklahoma
dtiab@ou.edu --- uptecsh@aol.com
3709 Windover Drive
Norman, Oklahoma, 73072, USA
Content
1. Hydraulic Fracturing: (+Video)
2. Interpretation of pressure tests in hydraulically
fractured wells using pressure and derivatives,
linear plots, semi-log analysis
3. Derivative Techniques: uniform flux, infinite
conductivity fractures
4. Finite conductivity fractures
5. Average Reservoir Pressure in HF reservoirs
6. Inclined Hydraulic Fractures
7. Average Reservoir Pressure
8. MHF and Fracture Tortuosity
9. Practical Aspects of Acidizing & Acid Fracturing
(+Video)
10. Numerical Examples
INTRODUCTION
5
σmin
σmin
σmin
9
σmin
σmin
10
σmin
σmin
11
16
GEOMETRY OF HYDRAULIC FRACTURES?
17
18
σmin
σmin
19
This situation is (on rare occasions) encountered in areas of
thrust faulting or where erosion has occurred after
deposition; in both cases rocks may be under greater
horizontal compressive stress than vertical overburden
stress.
σmin
σmin
20
Figure 1. Figure 2.
23
3.I -
CONVENTIONAL
TECHNIQUES
26
w
Well
Fracture
Thickness
Impermeable
boundaries
xf
27
C f = k f w f = 10000 kh (1.1)
w
Well
Fracture
Thickness
Impermeable
boundaries
xf
28
xf −s
rw ' ≈ ≈ rw e
Fracture
(1.2)
Thickness
2
Impermeable
boundaries
xf
conductivity fractures, xf
finite-conductivity fractures.
29
N o-flow Bounda ry
xf
xe
30
UNIFORM-FLUX FRACTURES
In some fractures, fluid enters the fracture at a uniform flow
rate per unit area of fracture face so that there is a pressure drop
in the fracture.
INFINITE-CONDUCTIVITY FRACTURES
Some fractures are assumed to have infinite permeability
(conductivity) and, therefore, uniform pressure throughout.
k f wf
C fD = FCD = 1.3
k xf
Xf = half-fracture length
Wf = fracture width
Kf = fracture permeability
Except for highly propped and conductive fractures, it is
thought that the uniform-flux fracture better represents reality
than the infinite-conductivity fracture, especially when the
fracture-face is damaged.
32
PSEUDO-RADIAL FLOW
(e)
33
FRACTURE FLOW REGIMES
1E+4
plot.
1E+1
Conductivity Fracture)
m = 1/4
1E+0
-- The occurrence of formation linear flow is 1E-7 1E-6 1E-5 1E-4 1E-3 1E-2
t, hr
1E-1 1E+0 1E+1 1E+2 1E+3
ELLIPTICAL FLOW
-- After a sufficient period of time has passed
flow begins to converge radially to the LINEAR FLOW IN THE FORMATION ELLIPTICAL OR BIRADIAL FLOW
25
INFINITE-
CONDUCTIVITY
&
UNIFORM-FLUX
FRACTURES
35
BASIC MATHEMATICAL MODEL
The dimensionless pressure drop of a well located at the center of a plane
vertical fracture can be generalized as follows:
1− x
+ erf 1 + xD − 0.25(1 − x )Ei (1 − xD ) − 0.25(1 + x ) Ei (1 + xD )
2 2
pwD = 0.5 πt Dxf erf D
1.4
2 t Dxf 2 t D 4t D 4t
Dxf Dxf Dxf
1 1
PD = π tDxf erf − Ei − 1 1.5
2 t Dxf 2 4tDxf
1 0.134
PD = πt Dxf erf + erf 0.866 − 0.067 Ei − 0.018 − 0.433Ei − 0.75 1.6
2 t Dxf t Dxf t t
Dxf Dxf
2
r
t Dxf = tD w = 0.0002637 kt
1.7
x φµct x f 2
f
36
A – FORMATION LINEAR FLOW IS DOMINANT
(UF Flux & Infinite Conductivity Fractures)
PD = π t Dxf
(b)
1.8
LINEAR FLOW IN THE FORMATION ELLIPTICAL OR BIRADIAL FLOW
(c) (d)
∆P = Pi − Pw = mvf t 1.9
− 4.064qB µ
mvf =
h kφct x 2f
37
Once the linear flow regime is identified from the loglog plot, a
Cartesian plot of dP versus time (Eq. 1.9) should yield a straight line
(which goes through the origin) of slope mvf which is then used to
calculate the half-fracture length Xf:
∆P = mvf t
4.064qB µ
xf = 1.11
hmvf kφct
38
162.6qµB
k= (1.12)
mh
and skin:
∆P k
s = 1.1513 1hr − log
2
+ 3.23 (1.13)
m φµc r
t w
39
∆ PSSL ≥ 2 ∆ Pel
40
Xf FROM LOGLOG PLOT
Let ∆PL1 be the value of ∆Pw at time t = 1 hour on the linear flow
straight line (extrapolated if necessary), then the Eq. 1.10 becomes:
4.064qB µ
xf = (1.14)
h (∆P )L1hr kφct
Thus, the fracture conductivity can be ∆PL1
calculated directly from the loglog plot
as long as the 0.25 slope is well defined.
41
EXERCISE 1
42
Pws, psi
Pressure buildup data obtained after a hydraulic ∆t, hr
fracturing treatment is given in Table. Reservoir 1170
0
characteristics are given below:
0.5 1340
q = 200 BPD rw = 0.198 ft 1 1395
h = 51 ft ct = 17.7x10-6 /psi 1.5 1452
tp = 364 hrs xe = 410 ft 2 1501
Ø=8% µ = 0.45 cp 3 1570
B = 1.507 bbl/STB tp=364 hours 4 1639
6 1748
Use the Conventional technique to 10 1899
calculate/estimate: 18 2075
1. Permeability (MDH plot)
27 2209
2. Skin
3. half fracture length 36 2304
45 2375
54 2434
63 2481
71 2521
43
44
SOLUTION MDH
MDH & Horner plots for the data
given in Table yield a late-time IARF m=741
straight line of slope = 741 psi/cycle.
∆P k
s = 1.1513 1hr − log + 3.223
2
m φµc t rw
− 100 0.58 × 106
= 1.1513 − Log + 3.223 = −4.9
2
741 0.08 × 0.45 × 17.7 × 0.198
46
4.064qB µ
xf =
hmvf kφct
( 200)(1.507) 0.45
x f = 4.064
(51)( 232) (0.54)(0.08)(17.7 × 10 −6 )
= 79.38 ft
48
4 - The fracture conductivity can be estimated from (Tiab, 1995):
3.31739k
wf k f = s
e 1.92173
−
rw xf
3.31739 × 0.54
w f k f == = 126.5 md − ft
e −4.88 1.92173
−
0.198 79.38
This equation will be discussed later in this chapter.
k f wf 126.5
C fD = = = 2.95
k xf 0.54 × 79.38
This value of CfD implies a low conductivity fracture.
49
5 – Wellbore Storage Coefficient
FINITE-
CONDUCTIVITY
FRACTURES
51
following equation:
LINEAR FLOW IN THE FORMATION ELLIPTICAL OR BIRADIAL FLOW
(c) (d)
2.45 / 4
PD = 1/ 2 t 1Dxf (1.15)
C fD PSEUDO-RADIAL FLOW
(e)
CfD
0.1
response equation is:
1.E+01
low conductivity fracture
DIMENSIONLESS WELLBORE
500
CASE I
44.13qµ B 1/ 4
I
∆P = ∆t = mbl ∆t 1/ 4
1.E-02 CASE II
m=1/4
L
f f t 1.E-04
1.E-10 1.E-08 1.E-06 1.E-04 1.E-02 1.E+00 1.E+02
DIMENSIONLESS TIME, tDxf
(1.16)
44.13qµ B 44.13qµ B
2
mbl =
h (k w )1 / 2 (φ µc k )1 / 4 k f w f =
( )1/ 4
(1.17)
f f t hm
bl φ µ ct k
52
Eq. 1.16 can be written as: (1.18)
The slope, mbl, of a straight line drawn through the data can be
used to determine fracture conductivity as follows:
2
44.13qµ B
k f w f = (1.19)
( )1/ 4
hmbl φ µct k
53
kf wf FROM LOGLOG PLOT
Let ∆PBL1 be the value of ∆Pw at time t = 1 hour on the bilinear flow
straight line (extrapolated if necessary), then Eq. 1.18 becomes:
54
2 2
44.13qµ B 44.13qµ B
k f w f = = (1.22)
h∆PBL1 (φ µ ct k )1/ 4
BL hm (φ µ ct k )1/ 4
55
The conventional semilog analysis, which applies for tDxf > 10, is used
to calculate permeability and skin:
162.6qµB 2
k=
k f wf =
mk 0.75
mh
3.68∆P BL1 (φ µ ct )1/ 4
2
44.13qµ B
k f w f =
h∆PBL1 (φ µ ct k )1/ 4
∆P k
s = 1.1513 1hr − log
2
+ 3.2275
m φµct rw
56
EXERCISE 2
57
58
∆t Pws ∆P ∆t0.25
0.4 2944.2 79.2 0.80
0.6 2952.7 87.7 0.88
1 2964.6 99.6 1.00
2 2983.4 118.4 1.19
4 3005.9 140.9 1.41
8 3032.5 167.5 1.68
10.1 3042.1 177.1 1.78
24.1 3094.0 229.0 2.22
48 3123.0 258.0 2.63
72.1 3146.0 281.0 2.91
97 3160.3 295.3 3.14
120 3171.3 306.3 3.31
144 3180.4 315.4 3.46
192 3194.7 329.7 3.72
240 3205.8 340.8 3.94
360 3225.9 360.9 4.36
480 3240.2 375.2 4.68
600 3251.3 386.3 4.95
720 3260.4 395.4 5.18
59
60
1 – Permeability (m=114.4)
∆P k
s = 1.1513 1hr − log
2
+ 3. 223
m φµct rw
70 17.5
= 1.1513 − Log 2
+ 3.223 = −5.0
114.4 0.12 × 0.55 × 0.0000155 × 0.333
3 – Fracture Properties (dPBL1=99.6)
2
44.13qµB
k f w f =
( )1/ 4
h∆PBL1 φµc t k
2
44.13 × 375 × 0.55 × 1.4
= = 7012 md − ft
1/ 4
23.5 × 99.6(0.12 ×0.55 × 0.0000155 × 17.5 )
2
44.13qµ B
k f w f =
h∆PBL1 (φ µct k )
1/ 4
64
MODERN
TECHNIQUES
65
1 – UNIFORM FLUX
FRACTURE
66
Fracture
Thickness
xf
PSEUDO-RADIAL FLOW
(e)
67
Fig. 2.1 shows the first published (Tiab, 1982) Type Curves of the
dimensionless pressure drop and the derivative for three values of
the ratio xe /xf of a well located at the center of a plane vertical
fracture in a bounded reservoir.
100
Pseudosteady state
No-flow Boundary
10
tDA
64
100
Pseudosteady state
10
P wD and t DA×P wD'
f= 8
xe/ x
1
f= 4
x e/ x
Radial flow
0.1 xe/xf=1
tDA
69
1 - For short producing times, fluid flow into the fracture from the
matrix is linear (i.e. formation linear flow regime).
Pseudosteady state
xe 10
xe/xf=1
Where:
0.1
kh ∆P
0.0001 0.001 0.01 0.1 1 10
0.0002637 kt
t DA =
tDA
and PD =
φ µ ct A 141.2q µ B
π xe
log(t DA × PD ' ) = 0.5 log t DA + log (2.2)
xf
70
Substituting for the dimensionless terms in Eq. 2.1 and solving for the
derivative of well pressure:
0..5
qB µ
m L = 4.064 (2.4)
h φ ct k x 2f
0 .5
2.032 Bq µ
xf = (2.6)
h ( t × ∆P ' ) L 1 φ c k
t
Note: Use this equation only if the FORMATION linear flow regime
is observed.
It is NOT applicable to the early fracture flow regime.
72
2 - The equation of the linear flow line portion:
70.6 qµB
k = (2.10)
h ( t ×∆P ') R
74
4 - The S factor is obtained from:
( ∆P ) R k tR
s = 0 .5 − ln + 7.43 (2.12a)
(t × ∆P ' ) R φµc r 2
t w
∆Ps ∆Ps
The FE for a buildup test is: FE = 1 − = 1−
P * − Pwf ( ∆t =0 ) ∆P *
Where P* is obtained from Horner plot. If P* is not available, Pavg
can be used instead.
SEMILOG PLOT OF dP AND (t*dP’)
The S factor can also be determined from the semilog plot of dP and
(t*dP’).
Using the time of intersection (tint) of the straight line of slope m and
(t*dP’)r gives:
kt int
S = 4.2166 − 1.1515 log
2
(2.12b)
φµc t rw
m = 2.303(t × ∆P ' ) R
m = 2.303(t × ∆P ' ) R
77
5 – SQUARE SYSTEM
For long producing times, the pressure derivative function yields a
straight line of slope=1.
t DA × P' D = 2π t DA
No-flow Boundary Pseudosteady state
(2.13) 10
f=8
xe/x
xf 1
f=4
xe/x
xe Radial flow
tDA
t × ∆P ' = t
qB 10000
(2.14)
∆Pr =1955
φ
∆PL1=710
t*∆P'
slope=0.5 tLRi=1.3
∆P
100 tRPi=234
( )
(t*∆P')L1=310
tr=54.8
Or:
10
4.27φ ct Ah
0.001 0.01 0.1 1 10 100 1000
time, hr
78
10000
The drainage area (in ft2) is: ∆Pr=1955
∆PL1=710
1000 (t*∆P')r=404
qB
t*∆P'
tPSS
A=
4.27φct h (t * ∆Pw' ) PSS
slope=0.5 tLRi=1.3
∆P
(2.15) 100
(t*∆P')L1=310
tRPi=234
tr=54.8
slope=0.36
10
0.001 0.01 0.1 1 10 100 1000
∆P
2 0.000527k pss
(2.16)
C = 2.2458(X /X ) exp 1 − t
A e f φµc A (t × ∆P') pss
t
pss
The exponential (exp) function makes this equation extremely
unreliable, i.e. any slight error in any parameter inside the exp-
function would yield an erroneous value of CA.
79
6 - RECTANGULAR SYSTEMS
t DA*P wD'
square
1
xe/xf=16
Radial flow
Figure. 0.01
0.0001 0.001 0.01 0.1 1 10
tDA
80
10
square
1
xe/xf=16
Radial flow
xe/xf=4
0 .5
qB µ 1
m CB = 8.128 (2.19)
h φ c kA
t 8
The straight line of slope 0.5 is the finger print a the two closest
parallel no-flow boundaries of the rectangular system, or channel.
81
Let (t×∆∆P')CB1 be the value of (t×∆
∆P') at time t = 1 hr on the straight
line of slope =0.5 (extrapolated if necessary). Thus from Eq. 2.18:
(t × ∆P')CB1 = mCB
We therefore can use Eq. 2.19 to solve for the drainage area A:
2
µ
8.128qB
A=
(2.20)
φkc h (t × ∆P ' )
t CB1
1
8
If PSS has been reached then use Eq. 2.15 to calculate A.
82
7 - The linear flow line and the infinite acting radial flow line
intersect at:
x 2f
t LRi = 1207φµ ct (2.17)
k
7544 φµ ct A 2
t LPi =
kx 2f
(2.18)
where A = 4xe2
83
9 - The radial flow line and the pseudo-steady steady state line
intersect at tRPI :
301.77φ µct A
t RPi = k
(2.19)
If A is known and PSS line is not observed (e.g. short test), then this
equation should be used to “locate” the PSS line of slope 1
2 ∆Pr=1955
t RPi ∆PL1=710
x 1000
= e
(t*∆P')r=404
(2.20)
t*∆P'
t LRi
xf
slope=0.5 tLRi=1.3
∆P
100 tRPi=234
(t*∆P')L1=310
tr=54.8
slope=0.36
84
EXERCISE 3
85
86
88
1. Permeability – Using the TDS technique:
70.6 qµB
k =
h ( t×∆P ') R
70.6×200×0.45×1.507
= = 0.55 md
51×322
2. Skin
( ∆P )R k tR
s = 0 .5 − ln + 7.43 =
(t × ∆P ' )R φµct rw
2
905 0.55×18×10 6
0 .5 − ln + 7.43 = −4.8
0.08×0.45×17.7×0.198
2
322
89
m = 2.303(t × ∆P ' ) R
m=741
kt int
S = 4.2166 − 1.15 log
2
φµ c r
t w
0.55 × 3.4
= 4.2166 − 1.15 log −6 2
= − 4 .8
0.08 × 0.45 × 17.7 × 10 × 0.198
90
3. FLOW EFFICIENCY
∆Ps − 3091
FE = 1 − = 1− = 2.35
∆P * 2280 m=741
4. Half-fracture length
0 .5
2.032 qB µ
xf =
h ( t × ∆P ' ) L 1 φc k
t
0 .5
2.032 ×200×1.507 0.45
=
51×131 0.08×17 .7×10 − 6 ×0.55
= 70 ft
91
5. Pseudosteady state
It is obvious from the plot that pseudosteady state has not been
reached (no late-time straight line of slope=1
92
Vertical fracture
A = 4 X e2 = 4 × 4002 = 640000 ft 2
Well
xf
xe
2
t RPi 228 400
2
x
= e = = = 32.5
t LRi
xf
7 70
93
2 – INFINITE
CONDUCTIVITY
FRACTURE
94
w
Well
No-flow Boundary
Fracture
Thickness
Impermeable
boundaries xf
xe
xf
100
Pseudosteady state
LINEAR FLOW IN THE FRACTURE
(a) BILINEAR FLOW
(b)
10
=
f1
x e/x 6
wD'
1
=f 4
×PDA
(c) (d)
xe/xf=2
0.1 Bi-radial flow
0.01
0.0001 0.001 0.01 0.1 1 10
PSEUDO-RADIAL FLOW
(e)
tDA
95
100
Pseudosteady state
t DA×P wD'
f =1
xe/x
conductivity vertical fracture inside a 1
f=4
xe/x Radial flow
square system.
xe/xf=2
0.1 Bi-radial flow
(a) the linear flow line of slope 0.5 0.0001 0.001 0.01
tDA
0.1 1 10
For xe/xf > 8, the linear flow regime The characteristics and
is almost non-existent, and the interpretation of the other three
elliptical flow line is observed first. flow regimes (linear, radial and
pseudosteady state) are exactly
For xe/xf < 8, it is the radial flow the same as discussed for the
line which disappears. uniform-flux fracture.
96
The equation of the elliptical flow line is:
0.72
x
(2.21)
t × P ' = 0.769 e t 0.36
DA D
x DA
f
Vertical fracture Well
xf
0.36 0.36 xe
qµ B k (t × ∆P')R k
C = 7.268 = (2.23)
Ell kh φµ c A 9.714 φµ c A
t t
x e
0.72
Thus, the elliptical flow line can be identified by its slope of 0.36.
It is important not to confuse this elliptical flow line with the linear
flow line because their slopes (0.5 and 0.36) are relatively close.
0.72
x
(t × ∆P' )Ell1 = 0.7699C Ell e
x
(2.25)
f
1.388
C Ell
xf = 0.694 x e
(2.26)
( t ×∆P ') Ell 1
58
=
(t × ∆P ')Ell1 F ra c tu re
Thickness
C Ell (2.28) Im p e rm e a b le
0.7699 b o u n d a rie s
xf
A=
φµ c khC
t Ell (2.29)
Vertical fra cture
W ell
1 / 0.36
( t × ∆P ' ) R
xf
k
=
xe
φµ c
t 12.617(t × ∆P' ) ELL1
99
TIME OF INTERSECTION 10000
∆Pr
∆PBR1
t*∆P'
straight lines is best used for verification t*∆P'BR1 (t*∆P')r
∆P
100 slope=0.5 tLBRi
purposes or for designing a pressure transient tr
tRBRi
100
EXERCISE 4
101
The pressure data given in Table were taken from Ref. 18, this corresponds
to a drawdown test in a highly productive fractured well. One of the objectives
of the frac-treatment is to achieve wfkf=10kh. The pressure derivative for this
test was estimated and it is also reported in Table. Other known reservoir and
well data are:
1. Calculate:
(a) permeability,
(b) skin and
(c) Half-fracture length
(d) Fracture conductivity
(e) Dimensionless fracture conductivity. Is the fracture conductivity high
or low?
(f) Did we achieve our objective?
2. Is the pseudo steady state flow regime present? If not, how long the test
should have been run to reach it?
102
t, hrs Pwf, psi t×∆P', psi t, hrs Pwf, psi t×∆P', psi t, hrs Pwf, psi t×∆P', psi
0.010 5180.5 9.2 0.23 5114.4 36.3 5.2 4914.2 91.7
0.012 5178.6 10.5 0.28 5107.5 38.7 6.3 4897.0 93.8
0.014 5176.6 11.7 0.33 5100.1 41.4 7.6 4879.5 95.6
0.017 5174.3 12.8 0.40 5092.2 44.2 9.1 4861.7 97.2
0.021 5171.9 14.0 0.48 5083.8 47.3 11.0 4843.7 98.5
0.025 5169.2 15.3 0.58 5074.8 50.6 13.2 4825.4 99.7
0.030 5166.2 16.6 0.69 5065.2 54.1 15.8 4806.9 100.7
0.036 5163.0 18.0 0.83 5054.9 57.8 19.1 4788.3 101.5
0.044 5159.6 19.5 1.00 5043.9 61.7 22.9 4769.5 102.2
0.052 5155.8 21.1 1.20 5032.1 65.6 27.5 4750.6 102.8
0.063 5151.8 22.7 1.45 5019.7 69.5 33.1 4731.6 103.3
0.076 5147.5 24.3 1.74 5006.5 73.3 39.8 4712.5 103.8
0.091 5142.8 26.1 2.09 4992.6 77.0 47.9 4693.3 104.1
0.110 5137.9 27.9 2.51 4978.1 80.5 57.5 4674.1 104.4
0.132 5132.6 29.8 3.02 4962.9 83.7 69.2 4654.9 104.7
0.158 5126.9 31.8 3.63 4947.1 86.7 83.2 4635.5 108.8
0.191 5120.8 34.0 4.37 4930.9 89.3 100.0 4616.2 120.2
103
104
70.6qµB
k= =
h (t × ∆P ' )R
70.6( 2000)( 0.3)(1.5)
= 12 md
50(105.5)
Step 4 - From the linear flow regime line of slope 0.5 at time t = 1 hour:
∆P')L1 = 97 psi
(t×∆
106
Calculate the half-fracture length from Eq. 2.26. Vertica l fra cture
W ell
xf
Converting the area from acres to ft2 gives: xe
0.36
(t × ∆P')R k
0.36
105.5 12
C = = = 16.5669
Ell 9.714 φµ c A 9.714 0.24 × 0.3 × 14.8 × 10 −6 × 3484800
t
1.388
C
x = 0.694 x Ell
f e ( t × ∆P ' )
Ell1
1.388
16.5669
= 0.694 × 933.4 × = 104.4 ft
61.7
107
Step 6 - Select any convenient time tR during the
infinite acting radial flow period and read the
corresponding value of ∆P:
tR = 48 hrs
∆P)R = 507 psi
(∆
507 12 × 48
s = 0.5 − ln −6
+ 7.43 = −4.85
2
105.5 (0.24)(0.3)(14.8 ×10 )(0.4)
Step 8 - Verification
The observed value of the time of intersection of the linear and radial
flow lines is approximately 1.2 hr. The calculated value is:
x 2f 2
t LRi = 1207φµ ct = 1207 × 0.24 × 0.3 × 0.0000148 105 = 1.2 hr
k 12
Since the observed and calculated time of intersection are approximately
equal, we can conclude that the values of k, xf and s are correct.
108
kt int
S = 4.2166 − 1.15 log
2
φµ c r
t w
12 × 1.1
= 4.2166 − 1.15 log 2
= − 4 .8
0.24 × 0.3 × 14.8 × 10 × 0.4
6
109
Step 9 – The fracture conductivity may be accurately determined from the
bilinear flow line (which is not observed in the figure) or ESTIMATED from
the following expression:
3.31739k 3.31739 × 12
wf k f = s
= −4.85 = 31376 md − ft
e 1.92173 e 1.92173
− −
rw xf 0. 4 105
k f wf 31376
C fD = = = 24.9
k xf 12 × 105
The fracture achieved the intended objective since the fracture
conductivity (31376 md-ft) is much greater (more than 5 times)
than the formation capacity kh:
10kh = 10 × 12 × 50 = 6000 md − ft
k f w f > 6000 md − ft
110
301.77 Aφµct
t RPi =
k
301.77×3484800×0.24×0.30×14.8×10− 6
= = 93 hours
12.05
3 – FINITE
CONDUCTIVITY
FRACTURE
112
w
Well
No-flow Boundary
Fracture
Thickness
Vertical fracture
Well
Impermeable
boundaries xf
xe
xf
PSEUDO-RADIAL FLOW
(e)
113
114
(A) PRESSURE
During the bilinear flow regime, the dimensionless well
pressure behavior is given by:
2.451
PD = t 0.25 (3.1)
C fD Dxf
In real units:
44.13 qµB
mBL = (3.3)
(φµct k )0.25 h k f w f
115
The subscript BL stands for bi-linear. Taking the
logarithm of both sides of Eq. 3.2 yields:
116
(B) DERIVATIVE
The derivative of the well pressure during the bilinear flow regime
is:
Note that:
(k f w f ) 2
k= t LBLi (3.11)
869.37φ µct x 4f
B - The time of intersection of the bi-linear flow line and the radial
flow line on the derivative curve is:
φµct
t RBLi = 1677 3
(k f wf )
2
(3.12)
k
118
Summary
119
If all three flow regimes (bilinear,
formation linear, and radial) are
observed during the pressure test
and they are well defined on the
pressure derivative curve:
70.6qµB
k=
h(t × ∆P ')R
kt R kt int
∆PR S = 4.2166 − 1.5 log
S = 0 .5 − ln + 7.43
( )
Or 2
t *∆P
' 2 φµc t rw
R φµ ct rw
∆Ps
FE = 1 −
∆P ∗
120
0.5
2.032 Bq µ
xf =
h ( t × ∆P ' ) L 1 φ c k
t
2
121.74 qµB
k f wf =
φµct k h (t * ∆P ' ) BL1
Or
2
1947.46 qµB
k f wf =
φµct k h∆PBL1
e s 3.31739k
xf = s
1.92173 − >0
e 3.31739k (3.13) r w k
− w f f
rw wf k f
2
121.74 qµB ( ∆Pw )
k f wf = s = 0 .5 R − ln kt R + 7.43
φµct k h (t * ∆P ' ) BL1
( t * ∆P ' w )R 2
φµ c t rw
122
(b) Bilinear flow is not observed:
3.31739k
wf k f = s
NOTE: Eq. 3.14 is valid only if:
e 1.92173 (3.14)
− e s 1.92173
rw xf − >0
rw xf
Where Xf is obtained from
linear flow line and k and s are 70.6qµB kt int
obtained from radial flow line: k= S = 4.2166 − 1.5 log
h(t × ∆P ')R 2
φµc t rw
0.5
2.032 Bq µ
xf = ( ∆Pw )
R − ln kt R
φ c k s = 0 .5 + 7.43
h ( t × ∆P ' ) L 1 ( t * ∆P ' w )R 2
φµ c t rw
t
123
(c) Radial flow line is not observed:
0.5
2.032 Bq µ
xf =
h ( t × ∆P ' ) L 1 φ c k
t
2
121.74 qµB
k f wf =
φµct k h (t * ∆P ' ) BL1
124
Note: Equations 3.13 and 3.14 are extremely sensitive to the
values of s, because of the exponential term.
They should be used only as a last resort and only to estimate the
half-fracture or the fracture conductivity.
EXERCISE 5
(Same data as Exercise 2)
126
∆t ∆P ∆t*∆P’
∆t Pws 1 99.6 24.99
1 2964.6
2 2983.4 2 118.4 29.76
4 3005.9 4 140.9 35.39
8 3032.5 8 167.5 40.53
10.1 3042.1 10.1 177.1 45.12
24.1 3094.0 24.1 229.0 49.85
48 3123.0 48 258.0 51.17
72.1 3146.0
72.1 281.0 51.63
97 3160.3
120 3171.3 97 295.3 50.41
144 3180.4 120 306.3 50.80
192 3194.7 144 315.4 49.68
240 3205.8 192 329.7 49.68
360 3225.9 240 340.8 49.68
480 3240.2 360 360.9 49.68
600 3251.3
480 375.2 49.68
720 3260.4
600 386.3 49.68
720 395.4
128
Solution
From the bilinear flow
line:
(∆∆Pw)BL1 = 99.6 psi
∆Pw′) BL1 = 25 psi
(t*∆
130
70.6qµB
k=
h ( t * ∆P ' ) R
(70.6)(375)(0.55)(1.4)
= = 17.5 md
( 23.5)( 49.68)
( ∆P) R kt R
s = 0.5 − ln + 7.43
(t * ∆P' )R
2
φµc t rw
1 375.2 (17.5)( 480)
= − ln −6 2
+ 7.43 = −5
2 49.68 (0.12)(0.55)(15.5 × 10 )(0.333 )
131
The skin factor can also be determined from the IARF lines on
semilog plot of dP and derivative.
kt int
S = 4.2166 − 1.5 log
2
φµc t rw
17.5 × 0.85
= 4.2166 − 1.5 log −6 2
= −5.1
0.12 × 0.55 × 15.5 × 10 × 0.333
132
2
121.74 qµ B
k f wf =
φµct k h (t * ∆P ' ) BL1
2
121.74 (375)(0.55)(1.4)
k f wf = = 7012 md − ft
−6
(0.12)(0.55)(15.5 × 10 )(17.5) ( 23.5)( 25)
133
1.92173 1.92173
xf = = = 166.5 ft
e s 3.31739k e −5 3.31739 × 17.5
− −
rw wf k f 0.333 7012
wf k f 7012
C fD = = = 2. 4
xfk 166.5 × 17.5
134
VERIFICATION
t RBLi = 1677
φµ c t
3
(k f w f )2
k
( 0 . 12 )( 0 . 55 )(15 .5 × 10 − 6 )
= 1677 3
( 7012 ) 2 = 15 . 85 hr
(17 . 5 )
Drainage Area,
Average Reservoir
Pressure, Shape Factor
136
1 2.2458 A
PwD (t DA ) = 2π t DA + ln +S (4.1)
2 C A rw2
1 4 x 2.2458 xe2
PwD (t DA ) = 2π t DA + ln +S (4.2)
2 C A rw2
xf
S = − ln (4.3)
2rw
137
1 35.933 xe2
PwD (t DA ) = 2π t DA + ln (4.4)
2 C A x 2f
PwD 1 35.933xe2
= 1+ ln (4.5)
(t DA xPwD ' ) 4πt DA C A x 2f
qB tPSS
A=
4.27φct h (t * ∆Pw' ) PSS
0.003314 k t pss (∆Pw ) pss
2
x
C A = 35.933 e exp 1 −
x φµ ct A (tx∆P ' ) (4.6)
f w pss
138
If the actual skin factor obtained from well test is not equal to the
result of Equation 4.3, then either Equation 4.8 or 4.9 should be used
to estimate the shape factor:
C A
Equations 4.11 and 4.12 are only valid when the assumption in
Equation 4.2 is valid, which is rarely the case.
140
1000
The Fig. shows a typical behavior Pseudosteady state line
100
system (see next slide). (t*∆Pw')pss
∆P and tx∆
3
4 0.098 0.9 0.6 0.015 1 0.1155 4.0 2.0 0.01
4
xf/xe=0.3
3.3351 0.7 0.25 0.01 1 1.9986 0.175 0.09 Cannot use
1
xf/xe=0.5
1 21.8369 0.3 0.15 0.025 1
1.662 0.175 0.09 Cannot use
1
1
xf/xe=0.7
1 10.8374 0.4 0.15 0.025 1 1.3127 0.175 0.09 Cannot use
1
1
142
EXERCISE 6
143
144
145
t Pwf t dP t*dP'
0.1 2450 0.1 50
0.25 2438 0.25 62 20.42
0.5 2420 0.5 80 23.80
1 2405 1 95 26.69
2 2383 2 117 32.46
4 2360 4 140 39.67
8 2328 8 172 45.14
10 2318 10 182 46.15
24 2273 24 227 50.09
48 2239 48 261 48.72
71 2220 71 280 49.21
96 2205 96 295 49.48
120 2194 120 306 49.33
144 2185 144 315 50.44
192 2170 192 330 48.01
240 2160 240 340 56.92
360 2128 360 372 89.69
480 2100 480 400 113.18
600 2072 600 428 140.94
720 2044 720 456
146
SOLUTION
The loglog plot of dP and derivative yields three straight lines.
3. The late time straight line has a slope=1 (PSS). Thus A can be
calculated.
147
70.6qµB
k=
h ( t * ∆P' ) R
(70.6)( 250)(0.4)(1.4)
= = 20 md
(10)( 49.4)
rw = 0.225 ft
h = 10 ft
µo = 0.40 cp
2. The skin factor is: Bo = 1.40
ct = 5x10-5 psi-1
( ∆P) R kt R Ø = 0.10
s = 0.5 − ln + 7.43 Swi = 0.20
(t * ∆P' )R
2
φµc t rw Pi = 2500 psia
q = 250 STB/D
1 280 ( 20)(71)
= − ln −5 2
+ 7.43 = −3.6
2 49.4 ( 0.10)( 0.4 )( 5 × 10 )( 0.225 )
148
2
121.74 ( 250)(0.4)(1.4)
kf wf = = 5295 md − ft
(0.1)(0.4)(5 × 10 −5 )( 20) (10)( 26.69)
149
4. Drainage area
From the PSS flow line at tpss=600, dPpss=428 and
(t*dP’)pss=140.94:
qB t PSS
A=
4.27φc t h ( t * ∆Pw ) PSS
'
4x e2 = A = 6.9788 × 10 6
X e = 6.9788 × 10 6 / 4 = 1321 ft
150
5
5. Since the linear flow regime is not Vertical-Fractured
observed, the half-fracture length is can reservoirs
CA
only be ESTIMATED from the correlation : xf/xe=0.1
1
2.6541
1.92173 1
xf = s xf/xe=0.2
e 3.31739k 1
2.0348
−
rw wf kf 1
xf/xe=0.3
1.92173 1 1.9986
= −3.6
= 18.4 ft
e 3.31739 × 17.5 1
− xf/xe=0.5
0.225 5295 1
1.662
1
4x = A = 6.9788 × 10
2
e
6
1
xf/xe=0.7
1.3127
X f / X e = 18. / 1321 = 0.014 1
xf/xe=1.0
1 0.7887
CA cannot be determined from
1
the table.
151
INCLINED
HYDRAULIC
FRACTURE
TWO UNIQUE PROBLEMS:
• Inclined Fracture
• Tortuosity
152
w
Well
Fracture
Thickness
Impermeable
boundaries
xf
154
INTRODUCTION
Most of the pressure transient analysis techniques to
analyze pressure responses of fractured wells are based
on the assumption that the fracture is either vertical (or
even horizontal).
156
REORIENTATION OF FRACTURE
158
Vertical HF Inclined HF
w
Well
Fracture
Thickness
Impermeable
boundaries
xf
159
160
1.E+01 1.E+01
Approximate Onset of Approximate Onset of
Pseudo-Radial Flow Early-Radial Flow Pseudo-Radial Flow
Early-Radial Flow
1.E+00 1.E+00
Bi-Radial Flow
P D & P 'D * t D
P D & P 'D * t D
Linear Flow
tD*P'D = 0.5 tD*P'D = 0.5
1.E-01 Linear Flow
1.E-01
0.68
tD*P'D = 0.5*(cos(θw))
tD*P'D = 0.5*cos(θw )
0.36
tD*P'D = 0.46*tD *cos(θw)
1.E-02 1.E-02
m = 0.5 m = 0.36
PD PD
1.E-03 1.E-03
1.E-04 1.E-03 1.E-02 1.E-01 1.E+00 1.E+01 1.E+02 1.E+03 1.E+04 1.E+05 1.E-04 1.E-03 1.E-02 1.E-01 1.E+00 1.E+01 1.E+02 1.E+03 1.E+04 1.E+05
tD tD
161
LINEAR FLOW REGIME
The pressure behavior of a uniform flux inclined fracture during the
linear flow regime is given by:
θw = Inclination angle of fracture
PD = cosθ w πt D 5.1 with well
50
In real units: 45
40
35
4.064qB cos θ w µ 1/ 2 θw
∆P, psia
30
∆P = t 25
h xf φ ct k
20
mLF
15
10
5
5.2 0
0 2 4 6 8 10 12
sqrt(t)
Eq. 5.2 indicates that a plot of ∆P
vs. t1/2 on a Cartesian graph will
yield a straight line during linear m = 4.064qB cosθ w µ
5.3
φ ct k
LF
flow with a slope: h xf
This slope can be used to calculate Xf/cosθw knowing k.
162
Taking the logarithm of both sides of 5.2
yields: 1000
(∆P)R=916.8
(t*∆P')L1=400
4.064qB cosθ w µ
log(∆p )L = log(t ) + log
1
∆t*∆P'
2 h xf φ ct k 100
(t*∆P')R=87
∆P
(t*∆P')ER =72
5.4
tR=800
time, hr
xf 4.064qB µ
= 5.5
cosθ w h(∆p )L1 φ ct k
163
From Eq. 5.2, the following equation for pressure derivative can be
derived:
1000
1 4.064qB cos θ w µ 1/ 2
(t × ∆p′)L
(∆P)R=916.8
= t (t*∆P')L1=400
2 h xf φ ct k
∆t*∆P'
5.6 (t*∆P')R=87
100
∆P
(t*∆P')ER =72
time, hr
solving for x f yields:
cos θ w
xf 2.032qB µ
= 5.7
cos θ w h(t × ∆p′)L1 φ ct k
164
EARLY RADIAL FLOW REGIME
This flow regime can only be observed in the loglog plot of derivative
curve.
The equation of the early radial flow is: (t × P ' ) = 0.5 cos θ D D ER w
In real units: 5.9
70.6qµB 1000
kh 5.10 (t*∆P')L1=400
= (t × ∆P ' ) R cos θ w
∆t*∆P'
(t*∆P')R=87
100
Solving for the angle:
∆P
(t*∆P')ER =72
(t × ∆p′) ER
cos θ w = 5.11
(t × ∆p′) R tR=800
LINEARFLOW
10
time, hr
Eq. 5.7, we can calculate Xf.
(∆P)R=916.8
70.6qµB
(t*∆P')L1=400
k= 5.13
h(t × ∆P ')R
∆t*∆P'
(t*∆P')R=87
100
∆P
Skin factor: 5.14 (t*∆P')ER =72
tR=800
kt R LINEARFLOW
∆PR
s = 0 .5 − ln + 7.43
( )
10
0.001 0.01 0.1 1 10
1 / 0.36
t
Note: Θ can also be estimated
cos θ w = LERi 5.11a
from:
t LRi
Where
tR = any time during the infinite acting radial flow line, and
∆PR = the value of ∆P corresponding to tR (from graph or table)
tLERi = intersection time of the linear and early-radial flow lines
tLRi = intersection time of the linear and infinite-acting radial flow lines
166
INFINITE-CONDUCTIVITY FRACTURE
1.E+01
will be discussed here. 1.E-04 1.E-03 1.E-02 1.E-01 1.E+00 1.E+01 1.E+02 1.E+03 1.E+04 1.E+05
tD
The equation for this straight line is: ( t D × p′D ) Ell = 0.46(t D0.36 ) cos θ w
In real units: 5.15
0 .36
3.34343qµµ cos θ w kt
(t × ∆p ′)ELL =
kh x 0f .72 φ µ ct
0 .36
5.16
cos θ k
= 0 .04737(t × ∆P')R 0 .72w t 0.36
xf φ µ ct
Thus the straight line corresponding to the elliptical flow regime has
a slope of 0.36.
168
( t × ∆p′) ELL1
0.36
cos θ w φ µ ct
= 5.18
0.72
xf 0.04737( t × ∆P ' ) R k
( t D × p′D ) ER = 0.5(cos θ w )
0.68
5.19 1.E+00
Bi-Radial Flow
P D & P 'D * t D
tD*P'D = 0.5*(cos(θw))
70.6qµB(cosθ w )
0.68 tD*P'D = 0.46*tD
0.36
*cos(θw)
(t × ∆p′) ER =
1.E-02
m = 0.36
PD
kh 5.20 m = 0.5 PD'*tD
= (t × ∆p′) R (cosθ w )
0.68 1.E-03
1.E-04 1.E-03 1.E-02 1.E-01 1.E+00 1.E+01 1.E+02 1.E+03 1.E+04 1.E+05
tD
169
5.20 and 5.18 can written as:
1 / 0.68
(t × ∆p′) ER
cosθ w =
(t × ∆p′) R
5.21
It can be shown that:
0.53
t
cos θ w = ERELLi 5.22
t RELLi
Where
tERELLi = intersection time of the bi-radial and early-radial flow lines
tRELLi = intersection time of the bi-radial and infinite-acting radial flow lines
0.36
k (t × ∆P' )ER
x0.72
= 0.04737 5.23
(t × ∆p′)ELL1
f
φµ ct
170
SUMMARY (Method 1)
70.6qµB
1000
(∆P)R=916.8
k= (t*∆P')L1=400
Permeability:
h(t × ∆P ')R
∆t*∆P'
(t*∆P')R=87
100
∆P
∆PR kt R
(t*∆P')ER =72
Skin s = 0 .5 − ln + 7.43
(
t *∆P ' )
R φµ ct rw
2
LINEARFLOW
tR=800
10
0.001 0.01 0.1 1 10
time, hr
k
0.5
(t × ∆p' ) ER (t × ∆p′) ER
x f = 0.02878 cos θ w =
Linear Flow: (t × ∆p′) R
φ ct µ (t × ∆p' ) L1
0.36
( t × ∆P ' ) ER
1/ 0.68
k (t × ∆p′) ER
ELLIPTICAL Flow: x 0.72
= 0.04737 cosθ w =
( t × ∆p′) ELL1 (t × ∆p′) R
f
φ µ ct
Permeability:
k= 900
800
mR h 700
∆t*∆P'
600
(m)R=200.4
500
Skin s = 0.5 ∆P1 hr − ln k
∆P
400
+ 3.23
φµ c r 2
300 ∆P R1hr=350
(m)ER=165.8
m R t w 200
100
(t*∆P')ER =72 (t*∆P')R=87
0
0.001 0.01 0.1 1 10
time, hr
Linear Flow: ( t × ∆ p ′ ) L 1 = 0 .5 ( ∆ p ) L 1
1000
(∆P)R=916.8
(t*∆P')L1=400
m ER = 2.303 ( t × ∆p ′ ) ER
∆t*∆P'
(t*∆P')R=87
100
∆P
(t*∆P')ER =72
0 .5
k ( m ) ER m
xf = 0 . 025 cos θ w = ER tR=800
φ µ ( ∆ p ) L1
LINEARFLOW
c t mR 10
0.001 0.01 0.1 1 10
time, hr
ELLIPTICAL Flow: k
0.36
m
1 / 0.68
( m) ER cosθ w = ER
x 0.72
= 0.057136
( t × ∆P ' ) ELL1 = 0.36( ∆P ) ELL1 f
φ µ ct ( ∆p) ELL1 mR
172
EXERCISE 7
173
φ= 0.12 (∆P)R=916.8
µ = 0.37 cp (t*∆P')L1=400
ct = 2.55x10-5 psi-1
B = 1.22 bbl/STB
∆t*∆P'
h = 14 ft 100
(t*∆P')R=87
rw = 0.25 ft
∆P
pi = 3000 psi (t*∆P')ER =72
tR=800
LINEARFLOW
Estimate 10
0.001 0.01 0.1 1 10
• Formation permeability
time, hr
• Skin factor
• Half fracture length
• Inclination angle
173
174
SOLUTION
(t* ∆P’)r = 87 psia tr = 800 ∆Pr=916.8 psia
(t* ∆P’)L1= 400 psia ∆PL1= (t* ∆P’)ER = 72 psia
kt R
∆PR
s = 0 .5 − ln + 7.43
(
t *∆P ' )
R φµct rw
2
916 .8 14 .5×800
= 0 .5 − ln + 7.43 = −3.93
87 0.12× 0.37 × 2.55×10 −5 ×0.25 2
0.5
k (t × ∆p' ) ER 14.5 72
x f = 0.02878 = 0.02878 = 18.5 ft
φ ct µ
−5
(t × ∆p' ) L1 0.12 × 2.55 ×10 × 0.25 400
2
EXERCISE 8
176
Given a pressure test data (graph) and the following information:
q = 322 STB/D
φ= 0.16
µ = 0.66 cp
ct = 1.25x10-6 psi-1
B = 1.35 bbl/STB
h = 27 ft
rw = 0.33 ft
pi = 3500 psi
Estimate
1. Permeability
2. Skin factor
3. Half fracture length
4. Inclination angle
Note: Bi-
Bi-radial = Elliptical.
177
SOLUTION
(t* ∆P’)r = 91 psia tr = 100 hr ∆Pr=736.6 psia
(t* ∆P’)ELL1= 240 psia ∆PELL1=580 (t* ∆P’)ER = 30 psia
kt R
∆PR
s = 0 .5 − ln + 7.43
(
t × ∆P
'
)R
2
φµ c t rw
1 / 0.72 1 / 0.72
k
0.36
( t × ∆p' ) ER 8.24
0.36
30
x f = 0.04737 = 0.04737 −6 2 = 6 ft
φ c t µ ( t × ∆p' ) ELL1 0.16 × 1.25 × 10 × 0.33 240
1 / 0.68
( t × ∆p′) ER ACOS (0.2) × 180 o
1 / 0.68
30
cos θ w = = = 0.20 ⇒ θ w = = 79 o
( t × ∆p ′ ) R 91 π
178
Using data in previous exercise and semilog plot of pressure and
derivative calculate k, S, Xf and θ.
179
From semilog graph : mr=209.6 mer=69.1 ∆P1=310,
tERBRi =0.003 tRBRi = 0.065 ∆PBR1=694 (from loglog plot)
162.6qµB 162.6 × 322 × 0.66 ×1.35
k= = = 8.24 md
mr h 209.6 × 27
∆P k
S = 1.1513 1hr − log + 3.2275
2
mr φµct rw
310 8.24 ×10 6
= 1.1513 − log + 3.2275 = −4.66
0.16 × 0.66 ×1.25 × 0.33
2
209.6
1 / 0.72 1 / 0.72
k
0.36
mER 8.24 69.1
0.36
X f = 0.057136
= 0.057136 2 = 6.0 ft
φ ct µ (∆p) BR1
−6
0.16 ×1.25 ×10 × 0.33 694
1 / 0.68
ACOS (0.2) ×180 o
1 / 0.68
m 69.1
cos θ w = ER = = 0.20 ⇒ θ w = = 79 o
mR 209.6 π
0.53
ACOS (0.2) ×180 o
0.53
t 0.003
cos θ w = ERRBi = = 0.20 ⇒ θ w = = 79o
t RBRi 0.065 π
180
COMPLEX PATTERN OF
HYDRAULIC FRACTURES
Deep-Acidizing in NFR Multiple hydraulic fractures
Near-wellbore
Far-field
181
MULTIPLE HYDRAULIC FRACTURES AND
TORTUOSITY
Tortuosity* is one of the biggest challenges for shale and tight gas
hydraulic fracturing treatments, leading to:
• High near-wellbore frictional pressure loss
• Premature screenouts
• Reduced treating rate
• And poor production results.
182
INTRODUCTION
Hydraulic fractures are initiated and kept open by
pressure.
Generally, the rock will fail at its weakest point and the fracture
will initiate there.
However, if the pressure continues to rise, additional fractures
may be formed because of changes in near-wellbore stress, as
shown in Fig. 1.
Later, the fracture will reach its natural direction, according to the
formation’s stress orientation.
184
Despite the artificial-stress environment generated around
the wellbore, hydraulic-fracture treatments tend to
produce a small number of larger fractures.
186
The important issue is that the loss of pressure is real and can
account for a substantial portion of the observed net pressure—
the total net energy available to propagate the fracture.
IDENTIFYING TORTUOSITY
The most widely used method to identify tortuosity is the rate-
step-down test.
188
TRADITIONAL MITIGATION
NEW SOLUTION
190
ACID FRACTURING
When the acid reacts with the carbonate formation, fines are
always released.
Then the acid is pumped down the fracture to etch the fracture
walls, which creates fracture conductivity.
When the acid contacts the walls of the fracture, the reaction
between the acid and the carbonate is almost instantaneous,
especially if the temperature of the acid is 200°F or greater.
Fig. 2 Fig. 3
The distance increases because, in a wide fracture, there is less
turbulence.
This results in less mixing as the live acid moves down the
fracture; therefore, the viscous and leakoff properties of
the fracture fluid should be controlled to maximize fracture
width.
When the acid etches the fracture walls, the resulting fracture
conductivity can be several orders of magnitude more
conductive than similar treatments that use water-based fluids
and propping agents.
End of Chapter 4