Vous êtes sur la page 1sur 11

Design of Acid Fracturing Treatments

B. B. Williams, SPE-AIME,Esso Production ResearchCo.


D. E, Nierode, SPE-AIME,Esso Production ResearchCo.

Introduction
Acid fracturing is a production stimulation technique
that has been widely used by the oil industry. In such
a treatment, acid or a fluid used in a pad ahead of
the acid, is injected down the well casing or tubing
at rates greater than the rate at which the fluid can
flow into the reservoir matrix. This injection produces
a buildup in wellbore pressure sufficient to overcome
compressive earth stresses and the formations tensile
strength. Failure then occurs. forming a crack (fracture). Continued fluid injection increases the fractures length and width. Acid injected into the fracture reacts with the formation to create a flow channel that remains open when the well is put back on
production.
To achieve reservoir stimulation, an acid fracturing treatment must produce a conductive flow channel
long enough to alter the flow pattern in the reservoir
from a radial pattern to one that approaches linear
flow. McGuire and Sikora conducted an analog simulation of the productivity of a fractured well that
serves as the basis for predicting the stimulation
achievable with vertical fractures. Their study indicated that the variables that determine stimulation
ratio are the ratio of fracture length to drainage
radius, L/r., and the ratio of fracture conductivity
to formation permeability, wkf./k. To design an acid
fracture treatment, therefore, ]t is necessary to predict the fracture geometry during the treatment, the
conductive fracture length, and the fracture conductivity created by acid reaction.

A number of authors have studied various aspects


of acid fracturing treatment design. Methods for predicting fracture geometry were first proposed by
Howard and Fast. Techniques that give improved
results have recently been presented by Kiel,3 and
Geertsma and de Klerk.4 Although these last two calculation procedures differ somewhat in formulation,
the resulting geometry predictions are in agreement.
Either procedure, therefore, can be used to predict
the dynamic fracture geometry in acid fracturing
treatments.
Acid reaction characteristics have been studied in
static reaction tests by several authors5-9 and design
procedures using data from these tests were proposed
by Hendrickson et al. Use of the static test to design
acid fracturing treatments is of marginal value since
the test models only the ratio of fracture area to
acid volume. An improved design procedure was presented by Barron et al., who studied acid reaction
by flowing acid through a channel between limestone
plates and derived a correlation to relate acid penetration distance along a fracture to treatment variables. The usefulness of the correlation is limited,
however, since the experiments were run in a smoothwalled fracture, at room temperature, and with the
fracture oriented in a horizonal plane. Smith et al.
studied acid reacdofi at ldg~l teinperahwes in a re=
action cell where reactive plates of limestone were
rotated through acid and noted the effect of velocity
on acid spending time and acid penetration. This

A model has been developed that accurately predicts acid penetration distance;
it allows the eflects of fracture geometry, acid injection rate, jormation temperature,
..-. L . ..-.
.h- ;--l
,JJ
;VI +ho +vontmonf
dosims
)?esu[ts
predkled
acid concentration, and ruim Lypc LU UG ,rwd4w-4
.1. ,Ibe .,-.,8
. . . . . -v
.. .. .
by the model can be used in modijying acid treatments to maximize the stimulation ratio.
JULY, 1972 -P

849

technique is of limited usefulness since it models only


the effects of acid flow without scaling the fracture
geomet~. More recently it has been shown that for
typical field conditions, acid reaction at the rock surface will be fast relative to the rate of acid transfer
to the surface. It was further shown that a model
developed by Tern112 can be used to describe acid
reaction when kinetics is limited by mass transfer.
To use this model, itis necessary to have data on the
rate of acid transfer to the fracture wall during flow
along the rough-walled fracture.
Broaddus and Knox have earned out experiments
to determine the conductivity resulting from acid
reaction with different formations and have shown
that conductivity is a function of formation type, acid
concentration, and contact time between acid and
---- . The ~~fidu~ti~jty ~hai w-~[~5ufi f~rn an acid
~uu~.
treatment cannot be predicted with certainty. If formation core samples are available, however, qualitative measurements can be made with procedures
developed by Broaddus and Knox.

fracturing treatments results in Eq. 1.21Use of this


equation is complicated since a numerical integration is required to evaluate the eigenvalues & and
eigenfunctions E.(q).

To simplify the use of the theory, the solution to


Eq. 1 is presented in graphical form as Fig. 1. This
figure allows the dimensionless position,

0.712Lai%.
i

at which the acid concentration reaches a desired


level, C/CO, to be related to the mass transfer Peclet
number, w~/24D,. With the exception of the effecm.vmmr.+a..
th. t WI,l C+
t;t,a
Ay:m- *WW.IJW,W,A.,
-_ffiP;.m+ u~n , th.=
.1. w ..lla..l~
..Je
p.ua..w
.W cl .1, -.
..1=.7.
be known to predict the acid penetration distance
are fixed by formation characteristics or by fracture
dimensions.
Mtimg Coefficient. The effective mixing
coefficient, D,, is an adjustable parameter that must
be chosen to allow the mathematical model for acid
reaction rate to agree with laboratory experiments
designed to accurately simulate acid reaction rate in
the field. If acid flow were laminar and acid transfer
were by ion diffusion alone, the effective mixing coefficient would be equal to the ionic diffusion coefficient for the acid of interest. The mixing coefficient
will normally be larger than the ionic diffusion coefficient since it will include acid transfer due to
secondary flow. Secondary flow can be caused by
changes in acid density with reaction, wall roughness,
turbulence, or a combination of these effects. When
acid velocity is below the critical Reynolds number
t,.wh,llant
finw
mivhto
nm-.iirc
ac a
for tra~~i$icn +fi

.1 -l.
. 11 ...
11.u....~
--.0
..0 .
result of secondary flow induced by density differences imposed by acid reaction. Flow near the wall is
vertically downward since acid in the boundary layer
contains high concentrations of calcium chloride. In
the center of the fracture, flow is vertically upward
since the fresh acid is less dense than the acid near
the wall. This circulation causes more reaction near
the top of the fracture and leads to the uneven reaction reported by Smith et al.*2 If acid flow is turbulent,
the rate of acid transfer to the fracture wall is :enhanced by eddies and the effective mixing coefficient
increases as flow velocity increases.
To obtain accurate values for the effective mixing
coefficient it is necessary to measure acid reaction
rates under conditions that closely simulate field conditions. In the experiments, therefore, it must be possible to measure the acid reaction rate of the fluid as
it flows between parallel rough walls of reactive rock,
with the fluid loss into the walls representing the fluid
loss from the fracture. The model must be oriented
to represent flow along a vertical fracture to allow
gravity forces to properly influence acid mixing.
Furthermore, the temperature and pressure at which
reaction occurs must be controlled to simulate reservoir conditions. Procedures used to obtain mixiig
coefficient data are presented in the Appendix.
Effective mixing coefficient data for the reaction of
Effective

Model for the Acid Fracturing Treatment


The stimulation ratio resulting from an acid fracturing treatment will depend in large part on the distance
from the wellbore that reactive acid contacts the walls
of the fracture created by fluid injection (hereafter
this distance is called the acid penetration distance).
The exact acid penetration distance during an acid
fracturing treatment cannot be accurately predicted.
When a pad of fluid is used ahead of the acid to
achieve a long, wide fracture, the maximum possible
penetration distance will be calculated assuming that
acid reacts in the fracture geometry existing at the
end of the pad; the minimum possible acid penetration distance will be the dynamic fracture length cal/...1-+-/4
bu, a,Gu

.-. . ...-...,.
caxiullllu~

L..+
UICLL

a.. :A
lluIu

1-..
lUSS

:
13

. . . . ..*.. -1A-A u..


VU1l LIU1lCXL
Uy

L.,.
L1lG

viscosity of reacted acid. These two conditions, which


define the limits of acid penetration distance, are
designated as the reaction rate limit and the fluid loss
limit, respectively. The actual penetration distance
probably varies between the maximum and the minimum distance during a treatment since acid reaction
produces worrnholes leading from the fracture into
the formation, which will eliminate fluid loss control
by the pad fluid.
Dynamic Fracture Geometry

Prediction of the dynamic fracture geometry during


fracturing treatments has been discussed by several
authors. z-4 In this paper, we shall use techniques
described by Kie13to predict fracture geometry. Fluid
loss during injection of the fluid pad is described
using equations and data presented previously.
Acid Reaction During Acid Fracturing Treatments
Mathematical Model. It has been shown that the reaction of hydrochloric acid at the fracture wall is
extremely rapid,13 indicating that the distance to
which reactive acid will move alorig a fracture is
governed primarily by the rate of acid transport to the
fracture wall and not by reaction rate at the surface.
For the assumption that reaction rate is infinitely fast,
the mathematical model for acid reaction in acid

850

JOURNAL OF PETROLEUM TECHNOLOGY

hydrochloric acid and limestone are presented in Fig.


irlat-:1.
m.a
in
~~~!~
~).
~~~
~.~~q
~Q~ffi.&~~
9 \Ub
4.
LbI1l~
Ulb
obtained in these experiments is called D= since the
acid reaction rate at the fracture wall is essentially
infinite, These data, which include variations in temperature, acid concentration, rate of fluid loss, flow
velocity, and fracture width, correlate well with the
scaling factor 2.67 WNR.*. A linear fit of Fig. 2 allows D= to be predicted using Eqs. 2 and 3. Mixing
coefficients predicted using these equations should be
valid for the reaction of any acid and rock during
an acid fracturing treatment, so long as the acid reaction rate at the rock surface does not restrict the
over-all reaction rate.

2
II

For N .,. w <3,750:


D~ = (0.69 + 0.00293
X

NRe*

10-5 sqft/min

w)

(2)

(3)

For N,., w >3,750:


10.3 + 0.00587 N.(,w)

D= = (

X 10- sqft/min

Wall roughness siemificantly increases the mixing


coefficient measured in the acid reaction tests. Data
in Fig. 2 show that the apparent mixing rate measured with a smooth fracture wall is always less than
the rate for a rough-walled fracture. This means that
design procedures that rely on acid reaction data
taken in a smooth-walled fracture will predict acid
penetration distances that exceed the real penetration
distance.
The reaction rate of hydrochloric acid at the fracture wall will not necessarily be infinite if the formation to be treated is not a limestone. When surface
kinetics is not infinite, the coefficient required to fit
experimental data with the theory that was derived
assuming reaction rate to be infinite will reflect the
SUrfaCekineIic !irni!. Mixinc
-..-...= coefficient
------------ data for the
reaction of hydrochloric acid and Kasota dolomite
(Fig. 3) illustrate the adjustment that is necessary
when the surface reaction rate affects reaction kinetics. For convenience, smooth-walled cores were
used in these experiments and the effect of wall roughness was inserted by increasing the mixing coefficient
by a predetermined factor; that factor was derived
from experiments with Indiana limestone in which
mixing coefficients for rough and smooth cores were
forced to agree. The corrected effective mixing coefficient for hydrochloric acid reaction with Kasota
dolomite is a function of temperature and can be
related to D~ by Eq. 4.
D,
=
D%

1 exp

2,445
[(

~ +1460
.


508
.

)1

.2

.4

DIMENSIONLESS

.6

.8

ACID PENETRATION

1.

DISTANCE

0.712 LaV hn

Fig. lAcid

penetration

distance along a fracture.

-:15

:!p::~20~
o

4000
8000
,
SCALING GROUP -2.67

16,000
WNR*O

Fig. 2Effective mixing coefficient for reaction


of HCI and Indiana limestone.

0.8
I

o.

(4)

The mixing coefficients required to model the reaction of acid with other types of formations should
follow the general behavior illustrated above. To
determine the reaction characteristics of acid with
other formations, it would be necessary to run dynamic reaction rate tests to obtain effective mixing
nnt -available.
CQeffi.Cien~
data. When COrSSare ---.--. --- , the
---JULY, 1972

.4 -

0.2
0

t
1

200

100

TEMPERATURE-0 F
Fig. 3-Effective
-f
w,

mixing coefficient

LIP I and
, ,.
s.

K..-+.
,. -.

for reaction

rlnlarnita
. ,.!..-.

851

TABLE

1FORMATION
PROPERTIES FOR EXAMPLE
LIMESTONE FORMATION

7,500

Depth, ft
Formation thickness, ft

50
50
07

Gross

Net
Fracture gradient, psi/ft
Permeability, md
Porosity
Youngs modulus, psi
Poissons ratio
Resarvoir Fluid Propetiles
Viscosity, cp
Density, lb/cu ft
Compressibility, psi-
Reservoir temperature, F
Resewoir pressure, psi

0.5
0.12
6.45 X 10
0.25
0.5
52
0.0001
200
2,500

data presented for a typical dolomite or a typical


limestone can be used. The reaction time for the
formation rock in a standard static reaction test* may
be used to determine whether the reaction rate behavior is similar to that observed with Indiana limestone or to that observed with Kasota dolomite, Eqs.
2 through 4.

Example Treatment Design


The model for acid reaction in a fracture, in conjunction with a fracture geometry model, can be used to
design acid fracturing treatments. A proposed design
procedure is outlined below.
1. Select appropriate fluids for a pad to be injected
ahead of the acid. These fluids should be viscous at
reservoir temperature and should allow fluid losses
to the reservoir to be controlled.
2. Using one of the techniques described in the literature,- predict the dynamic fracture geometry
created by injection of the pad fluid.
~. ~l~ing the techniques described here, predict
the distance that acid can penetrate along the fracture
created by the pad fluid. Predictions should be made
for different pad volumes and acid injection rates in
order to determine the design that will give maximum
stimulation at minimum cost.
4. Specify the acid volume.
5. Predict the stimulation ratio for the treatments
of interest.
6. Select the treatment that will best satisfy company economic criteria.
To illustrate, let us design an acid fracturing treatment for a well completed in a limestone formation at
a depth of 7,500 ft. This formation has a conductivity
of 25 md-ft. with a net interval thickness of 50 ft. The
well produces an oil that at reservoir temperature
/~nnoE\

(Luu

1)

h..

,Ias

. .,;

ea. it.,

a v&u...J

fif
u.

(l < t-m f)thtar


v.a WY. v...-.

rt=ct-rvnir
nro. . . . . . . . . r.-

perties are summarized in Table 1.


1. Select

AppropriateFluids

A treatment will be designed using Pluid A (a fictitious viscous fluid) as a pad fluid. We will assume that
this fluid is viscous at reservoir conditions (about 60
Standard teat: auapend a l-cm cube of rock in a 25-cc gmduate
and add 10 cc of 15 percent HCL Allow the reaction to procaed for
5 minutes.at room temperature and atmospheric preaaum. Maaa.
ure the final acid concentration. In this teat, the final mmposltion
...-.
-a

852

= s - ---+
4.- per..-!!.

!!

h-

. ..-

l;-m.+ap
-m-l 10 A nnment
. . . . .. . ..- .- ------ . ~-------

f~r fhe

dolomite.

TABLE

2TREATMENT
CHARACTERISTICS
FOR THE
EXAMPLE LIMESTONE FORMATION

pad Fluid (Fluid A)


Maximum Permissible Injection Rate (set by
sutiace pressure and tubing size), bbl/min
10
Temperature at which fluid enters the fracture, F
150
cn
w
Average viscosity during fiaw a!crlg the free%iie, Cp
Fluid loss additiva concentration, lb/1,000 gal
20
Fluid Loss Characteristics
Spurt volume, gal/sq ft
0.007
Fluid loss coefficient (C~), ft/min%
0.002
Acid
~mum
Permissible Injetilon Rate (set by surface
prassure limit and tubing size), bbl/min
20
Average Viscosity for Flow Along the Fracture
(175F, patially raacted acid), cp
15 percent HCI
0.8
28 percent HCI
1.2
Viscosity of the Reacted Acid (200 F)r cp
15 parcant HCI
0.8
28 percent HCI
1.7
Acid Density, lb/cu ft
15 percent HCI
66.8
28 parcent HCI
71.1

cp at shear rates for flow along the fracture) and can


be pumped through well tubular goods at 10 bbl/min
without exceeding the allowable surface pressure of
5,000 psi. To reduce the rate of fluid loss to the formation, Fluid A contains a fluid loss additive at a
concentration of 20 lb/1,000 gal. Both 15- and 28percent hydrochloric acid (HCl) will be considered
for the treatment. If a friction reducer is added, these
acids can be pumped at a maximum rate of 20 bbl/
min. The average viscosity of the acid during flow
along the fracture (assuming that the temperature
averages 175F* * and that 50 percent of the acid has
reacted) is estimated to be 0.8 cp for 15-percent HC1
and 1.5 Cp fix 28-pcerlt
He:. The viscosity of
totally reacted acid as it flows into the formation
perpendicular to the fracture (at 200F) is estimated
to be 1.0 cp for 15-percent HC1 and 3.0 for 28percent HCL Other fluid properties are summarizal in
Table 2.
2. Predict the Dynamic Fracture Geometry
The dynamic fracture geometry created by the pad
fluid was predicted using equations disclosed in prior
publications. g This calculation was made for data
given in Tables 1 and 2 for the assumption that the
fracture is vertical and has a constant vertical height
of 50 ft. Results are summarized in Table 3 for various times during the injection of the pad fluid. The
geometric factors included in the table indicate the
expected fracture size and shape after injection of
volumes of pad fluid ranging from 150 to 600 bbl.
3. Predict the Acid Penetration Distance
The penetration distance (defined as the point where
C/Co n O.1) is shown in Table 4 for 15 and 28 percent HC1 injected at 10 and 20 bbl/min. This calculation involves the following steps:
a. Calculate average fluid loss velocity along the
preditiedwithtechniques described by slrlchir. is

JOURNAL OF PETROLEUM TECHNOLOGY

TABLE 3-~ACIURE
BY THE

GEOMETRY
-. .. . . . CREATED
PAD FLUID

(Example
Fluid
Volume
Injected
(bbl)
~~~

Formation)

Tima
(rein)

]~
30
45
60
Length

Limestone
Average
Fracture
Width
_ (in.)
&:4

Fracture;
Volume
(w ft)
~~~

281
348
405

416
600
780

0.18
0.21
0.23

300
450
600

*Volume

Fracture
Length
_ (ft)
~.~

of one wing of a rectangular-vertical


of both wings of the fraCt We.

d. Read the effective mixing coefficient from Fig.


2 or calculate a value using Eq. 2.
e. Caicuiate the Peciet number for the treatment,
using Eq. 8.
Npe =

fracture.

;.-

2\/t

UIC

~LLcllll

* = 33.68 i
,
0

ft/min

(6)

c. Calculate the scaling group, 2.67 wNR,., used


to correlate the mixing coefficient data.
TABLE 4-CALCULATION

Time
(~~n)

0.0008

15
30
45
60

48.2
37.4
32.0
29.2

0.00057
0.00047
0.00041

ACID

15 parcentHCI
10 bbl/min
20 bbl/min

15
30
45
60

1.64
1.91
2.12
2.25

Time
(rein)

2.60
3.14
3.54
3.82

96.4
74.8
64.0
58.4

0.30
0.24
0.21
0.19

0.19
0.15
0.13
0.12

TABLE 5-SUMMARY

Pad Volume
(bbl)

150
300
450
600
JULY, 1972

0.38

0.25

0.18
0.16
0.15

OF CALCULATED

Maximum Acid
15 percent HCI
10 bbl/min
20 bbl/min

105
118
125
130

133
148
155
164

1,740
2,232
2,600
2,848

0.28
0.22
0.19
0.17

1.96
2.32
2.59
2.77

0.30
0.26
0.24

DISTANCE

0.18
0.14
0.12
0.10

1,160
1,488
1,733
1,899

28 percentHCI
20 bbl/min

0.35
0.28
0.25
0.23

0.24
0.18
0.16
0.14

MaximumAcidPenetrationDistance(ft)
15 percentHCI
28 percentHCI
20 bbl/min
10 bbl/min
10 bbl/min
20 bbl/min
105
118
125
130

RESULTS,

Penetration

580
744
867
949

Peclet Number
15 percentHCI
10 bbl/min
20 bbl/min
lin

Dimensionless Acid PenetrationDistance


15 percentHCI
28 percentHCI
10 bbl/min
20 bbl/min
10 bbl/min
20 bbl/min

15
30
45
60

PENHRATION

870
1,116
1,300
1,424

28 percentHCI
10 bbl/min
20 bbl/min
1.33
1.51
1.64
1.73

(9)

Scaling Groupfor Flow Along the Fracture


15 percent HCI
28 percent HCI
.m L., ,:
a,-, . . . ,,
,m Il., #:_
mmL., ,:LU Dm{man
LU uul[mtn
LU Dmlmln
zu mfmm

Mixing Coefficient(105 x Dm, sq ft/min)


Time
(rein)

lv~l

Results from the acid penetration calculations are


summarized in Table 5. Included in this table is the
maximum penetration distance for both 15- and 28percent HCI injected at 10 or 20 bbl/min and the
dynamic fracture length when acid injection was
started. Penetration distances are listed for pad volumes of 150, 300, 450, and 600 bbl of Fluid A. Four
treatments that appear to have the greatest potential
stimulation ratio have been selected for further analysis. These treatments are described in Table 6. In
actual practice, this technique would be programmed
for a computer so that the economic potential of all
cases could be analyzed.
The procedure just outlined should now be re-

OF MAXIMUM

Add FlowVelocity(ft/min)
10 bbl/min
20 bbl/min
Injection
Injection
;a;e
ia~e

Average
FluidLOSS
Velocity
!$+l... \..\
(,,1*,,,111

m,

AA

1 .+U4

;hn

(5)

This procedure should be carried out for the fluid


. . .
fl..:.-l
/+,.
-++-:..
L-..-:
1-CO&-uG1ll&lG1l
..--ffi,.:a..+L IU1
<-. +1.lllclA11u33
UK
~CIU
llUIU
{W
mum possible acid penetration distance) and for the
fluid loss coefficient for the acid alone (to estimate
the smallest possible penetration distance). For simplicity in this example, we will consider only the
maximum penetration case.
b. Calculate the flow velocity at the inlet to the
fracture using Eq. 6.
hgw

La =
,ft/min

(8)

f. Read the dimensionless acid penetration distance, N.A,from Fig. 1.


g. Calculate the acid penetration distance from
the definition for the dimensionless distance, Eq. 9.

fracture at each time of interest (i.e., pad volume)


using Eq. 5.14
7C,G

v
m

133
148
155
164

EXAMPLE
f)istance

LIMESTONE

133
148
155
164
FORMATION

(ft)

28 percent HCI
)0 bbl/min
20 hhl/min

133
148
155
164

176
177
191
205

176
177
191
205

Fracture
Length at the
End of the Pad
(ft)

194
281
348
405
853

TABLE 6-MAXIMUM

Treatment
Number
1

2
7

Treatment
300bbl pad at
5,500 gal of is
at 10 bbl/min
300-bbl pad at
8,000 gal of 15
at 20 bbl/min
Ztnn,

hhl n t-i a+
,
~a
-.

STIMULATION

450-bbl

EXAMPLE

LIMESTONE

FORMATION

Maximum Stimulation
Ratio,** J/J
~,

Maximum Acid
Penetration
Distance (ft)

Dimensionless
Fracture*
Length L./r.

118

0.18

2.0

3.7

4.1

148

0.22

2.0

3.9

4.6

~q

0.22

2.0

3.9

4.6

177

0.27

2.0

4.3

5.0

~.

~.

10 bbl/min
percent i-iCi

lfl
-

10 bbl/min
percent HCI
hhl/min
-.
, . . ... .

3,500 gal of 28 percent HCI


~~ ~~ ~~!,/m~R

RATIO,

pad at 10 bbl/min

7,000galof 28 pemiit i-m


at 20 bbl/min

*Assumes 40-acre well sDacinfz, r, = 66o ft.


** FrcIm curves published by M;Guire
and Sikora.l

peated for the assumption that fluid loss from the


fracture is controlled by the reacted acid viscosity.
To do this. first caicuiate the fluid kiss ~dfkiefit,
C,,., fQr the spent acid and then repeat the calculations to predict the acid penetration distance. The resulting acid penetration distances will represent the
minimum possible acid penetration distance to be expected from the treatment. Field results will normally
be between this minimum and the maximum vaiues.
Since the calculation of minimum acid penetration is
comparable with that outlined in Table 4, we will not
show it here.
4. Select the Acid Volume
The acid volume required to achieve a desired fracture conductivity can be qualitatively predicted using techniques proposed by Broaddus and Knoxs if
core samples are available. In many cases, cores are
not available and an approximate procedure must be
used to select the acid volume. We propose that if 28percent HCI is used the acid volume should be at
least 1.5 times the volume contained within the fracture between the wellbore and the maximum acid
penetration distance. If 15-percent HCI is used, the
acid volume should be about three times the fracture
volume. If, for example, the average fracture width
was 0.24 in., the height was 100 ft, and the acid
penetration distance was 150 ft, the fracture volume
in the region to be etched would be 4,500 gal (7.48
gal/cu ft X 2 wings X 150 ft/wing X 0.24 in./ 12
in./ft X 100 ft). At least 6,750 gal of 28-percent
HCI would be used in this treatment (4,500 gal X 1.5
= 6,750 gal). Acid volumes specified for the treatments considered in Table 6 were calculated using
this procedure.
5. Predict the Stimulation Ratio
The maximum stimulation ratio for the four treatments of interest is given in Table 6. These data were
predicted for a 40-acre well spacing using published
correlations, for the assumption that the acid will
etch a conductive flow channel throughout the area
are =..
~iven
~~n~a~ted. .Ctimlllatirnn
--- for a ratio of
...... .. ... .. .ratio~
. ... . --fracture conductivity to formation permeability of
10, 10, and 10 md-in./md to cover the range of
possible etched fracture conductivities.
A shortcoming of this or any other design proce854

dure for acid treatments is our inability to predict accurately the fracture conductivity to be created by
acid etching. wh~m
~nr~~ ~r~
can be
--------- , tests
-., .Jwl.------ available.
run to evaluate conductivity qualitatively. Unfortunately, these tests are often questionable. Fieid results are therefore the only realistic test for the effectiveness of a given design.
Treatment
~. .,.WW.
%lant ~bm
. .- Mnd
...-. lhmn~~
----.Treatment 4 appears to be the best of the four treatments considered in Table 6 since it will offer the
maximum potential for stimulation. An economic
analysis of this or other treatments should follow
general practices for evaluating workover or stimulation candidates. Technictues for economic analysis
will not be discussed her:.

Iw

= 0.1 INCH
hn

500

::
t

0.0005

T = 200

FT/MIN
F

o
INJECTION

- i/hg (BPM/FT)

Fig. 4-Effect
of injection rate on
acid penetration distance.

JOURNAL

OF PETROLEUM

TECHNOLOGY

Effect of Treatment Variables


On Acid Penetration
To illustrate how changes in treatment design could
modify the acid penetration distance, the effect of
several important variables will be considered for
the example well just discussed. These variables include the acid injection rate, the fracture width
created by the pad fiuid, the temperature, and the
acid concentration. The effect of these variables is
illustrated in figures in which all other parameters
are held constant.
Increasing injection rate will increase acid penetration, as shown in Fig. 4. At injection rates exceeding about 1.0 bbl/min/ft, the penetration distance
will approach a maximum of 300 ft for reaction with
limestone and about 450 ft if the formation is a dolomite. This maximum occurs at small Peclet numbers
and high Reynolds numbers where the dimensionless
acid penetration distance is linearly proportional to
the Peclet number and the effective mixing coefficient
is linearly proportional to injection rate.
.Aoirl
m-..,

flnw
1, ,,

rat,=
i/h ,.,
.-. W, .,

man
. . . . he
w

in~n=aced
1..
-...-..

hv
j

iniPI-tino
l.., ww.l..~

at a higher rate or by designing the treatment so that


acid contacts only a fraction of the total fracture
height. As proposed by Graham et al., the area conby using a viscous
tacted by acid can be restricted
pad fluid so that acid channels through the fluid rather than uniformly displacing it. In treating the example well, restricting the acid flow to one-fourth the
total fracture height (Fig. 5) could significantly increase acid penetration. if the injection rate is 0. i
bbl/min/ft (based on total fracture height), penetra-

tion can be increased from 110 to 265 ft in a dolomite or from 75 to 180 ft in a limestone.
The effect of fracture width on acid penetration
distance is illustrated in Fig, 5. In this example, an
increase in width from 0.1 to 0.2 in. increases acid
penetration distance from 120 to 177 ft in a limestone
and from 177 to 255 ft in a dolomite. The importance
of preceding acid injection with a viscous fluid pad is
evident since the fracture width w~l increase in proportion to the fluid viscosity raised to the 0.25 power.
The temperature at which acid reaction occurs
will affect the depth of acid penetration. As shown in
F]g. 6, an increase in temperature from 100 to
220F could decrease the penetration for 15-percent
HCI from 120 to 82 ft in a limestone and from 285 to
120 ft in a dolomite. The reduction in penetration
distance occurs because of the decrease in acid viscosity with temperature and the associated increase
in the effective mixing coefficient. To account for the
effect of reaction products (CO, and calcium chloride)
on the acid viscosity, it was assumed that the average
acid viscosity in the fracture was 1.25 times the visrncitv
w
.J

nf
.

<-m=rm.nt
.1 J
. . ..
y-. -w.. . .UCI
. . . and

7 0
-.

tim~c
. . . ..V

th,-...

vicrncitv
, .=.,..J

of 28-percent HC1.
In calculating acid penetration distance for field
treatments, it is important to predict the temperature of the acid as it enters the fracture and its average temperature in the fracture. The stimulation ratio
achieved by a treatment can sometimes be improved
by the use of a pad to reduce acid temperature in the
fracture. In the treatment of a dolomite (Fig. 6),
however, a pad that wouid reduce reaction temperature from 220 to 150F would increase the pene-

1
300

i = 10 BPM
% =hn=50Fl
- = 0.0005 FT/MIN

w = 0.1 IN

250
200 -

50

150 = 10 BPM
I

hg=hn=50FT

C/C.

FRACTURE
Fig. 5-Effect

-v = 0.0005
T = 200

FT/MIN
100-

= 0.1 (28%

HCI)

I
100
WIDTH

JULY,

1972

140

160

TEMPERATURE

- IN

of fracture width on
acid penetration distance.

120

Fig. 6-Effect

180

200

220

- F

of temperature and acid concentration


on acid penetration distance.
855

~ 1000

0.1

0.5
lN!E2aON

:A3E

0.6

- BP&l

Fig. 7<omparison
of acid penetration distance
predicted by the proposed model and
by a static reaction test.

1.00 :

II

,;

1!
11,

/
~

~-r

---

II
II

,,

11
//-

)--------74

#E/c_&~

ii

[ ! i

/
/

----i+

/1

tration distance of 28-percent HC1 from 177 to only


222 ft.
An increase in acid concentration from 15- to 28percent HC! wi!! increaw acid penetration di~$afi~e
since the more concentrated acid is more viscous and
will have a lower value for the effective mixing co-P~~ -e-e+--+
;-- di~~a~~e
e.ffi.~ie~-~.
-Fig. ~ ~how~ .,,kn*
-. au,u
pen L1clL1ull
in a 200F dolomite formation can be increased from
127 to 178 ft by using 28-percent HC1 instead of 15percent HCI. Penetration distance could be increased
further by adding materials that would increase the
acid viscosity and thereby reduce the effective mixing
coefficient. The value of additives that increase acid
viscosity can be estimated through the use of Eqs. 2
through 4 and Fig. 1.

Comparison of Techniques for Predicting


Acid Reaction in Fracturing Operations
The design procedure proposed in this paper d~ers
from previous ones in that it attempts to include all
the variables that can affect acid penetration distance.
The following discussion illustrates how this model
differs from procedures based on the static reaction
rate test, on flow experiments, and on combinations
of theory and flow experiments.
Design Based on Static Reaction Rate Test

The first procedure proposed for the design of acid


fracturing treatments was to use the acid spending

..01=2
tim.a
..I..

1000

ACID PENEIRATION

DISTANCE - FT

Fig. %Comparison
of acid penetration distance
predicted by the proposed model and by
the model of Barron et al.

+
:

175

Cvc = .007

150 ~

~
--b

GAL ---

I
1

tim
\

100
...

I
!

75

;
I
~

50.2

.4

.6 .8 1

248

AC!D EFFECTIVENESS !WMBE!?


Fig. 9-Comparison
of acid penetration distance
P~e~~~ted ~i the Pi@e~ed ~l~~~i ~fid b!!
the model of Whitsitt et al.
856

~10,000

.-.

FT/~N

G
~

E
Q
v

IN

CONC. = 20%

z
O 125

+
g

.05

i = 25 BPM

w=

A.af,arm;nmA

UG.GII...LIGU

f.fi!IUIII

.
a

.+..:
sLa L1&

--.
-.:-lCm,LIU1l

----

LCS1

.LU

-..:

CSLl-

mate penetration distance. In such a test, a sample


of formation rock is contacted by acid and acid concentration is measured as a function of time. Rock
surface area and acid volume are scaled to represent
the area-to-volume ratio expected in the fracturing
treatment. The time required for acid to react to 20
percent of the initial value is normally reported as the
reaction time for the acid-rock system. Penetration
distance is computed by determining the average
nlnn*L.=
:.. +kacid flow ~ela.-it.~
L1l G +---11 Gall+---+
llC1ll
,uw,ty a,,,~
,,,L f---+,
, clb Lu1-=b 111
and then computing the distance acid can travel during the spending time.
Typical reaction time data are given in Table 7.
In Fig. 7, acid penetration distance predicted using
these data is contrasted with the penetration distance
predicted by the proposed theory. * Predictions based
on the static reaction test are normally more optimistic than predictions from the proposed model since
the static test cannot reflect the effects of acid flow,
channel roughness, or fluid loss on acid penetration
distance. A design based only on static reaction rate
data can therefore be misleading.
Design Based on Flow Tests

Barron
Pt al. have presented acid design curves
based on scaled flow experiments. In their studies, a
flow model scaled to represent a fracture was used to
measure acid reaction. Experiments were run over
a wide range of fracture widths and acid flow rates.
Predictions based on these experiments are presented
Penetration
distance
f mm static test is calculated
assuming
that either L.= v,,,., t. or L. = v,,,.. t./2. In these equations, the maximum flow velocity along the fracture is taken to be the acid velocity
entering the fracture.

JOURNAL

OF PETROLEUM

TECHNOLOGY

in Fig. 8, where the acid penetration distance is related to injection rate and fracture width. Comparison
with our proposed technique shows that the penetration distance predicted by 13arron et a!. is !arger. ,+%
an injection rate of 0.1 bbl/min/ft and a 0.05-in.
fracture width, Barrons design curves predict an acid
penetration distance of 140 ft, as compared with 42
ft predicted with the proposed model.
The difference between models for acid penetration can be related to the experiment chosen to represent the acid fracturing process. Barron measured
acid reaction rate in a h&~zontal fracture model with
sirmoih limestone waiis. & shown in Fig. 2, the effective mixing coefficient in the smooth-walled fracture will be smaller than in the more realistic rough
fracture used in our study. The horizontal fracture
orientation will further reduce the effective mixing
coefficient by minimizing gravity effects.
Design Based on a Combination of
Theory and Scaled Experiments

Whitsitt ef al.; have reported a studv of acid reaction


. . . . . .:.
Uperttuoiis
h
which they obtained reiii fiWtitiihg
action rate data for reaction in flow between parallel
walk of limestone and other rocks. Experiments were
run in the temperature range of 70 to 150CF at a pressure of 1,000 psi and for 0.05- and O.l-in. fracture
widths. A mathematical model that incorporates data
TABLE -EFFECTIVE

i*/tt

(bbl/min/ft)

(ft/min)

0.05
0.06
0.06
0.O6
0.08
0.35
0.67
0.80
0.07
0.37
..-.
U.51
0.74
0.88
088

0.004
0
0.002
0.003
0.004
0
0
0
0
0
----

1.02
1.37
0.68
0.62
0.51
0.68
0.52
0.60
0.48
0.06
0.52
0.60
0.23
0.55
0.66
CL29
0.52

0.012
0
0.006
0
0
0
0
0
0
0.014
0
0
0
0
0
G
o

U.ul1

0
0.011
(-j~~y

RL rough fracture
SL smooth fracture
SD smooth fracture

JULY,

1972

Temperature
(F)

70
70
70
70
70
70
70
70
180
180
180
180
180
. -1au
180
183
184
70
167
70
145
70
192
70
140
187
70
70
70
--1 7(J
222

TABLE 7+EACTION
15 PERCENT
Pseudo Fracture
W!dth
(in.)

80 F

0.1
0.24

14
29

TIME
HCIG

FOR

Reaction Time (minutes)


150 F
200CF
3
12

7
29

for acid reaction was then derived, This work has not
been published in a form that makes the data accessible; however, typica! resu!ts have been pub!ishe&1
Fig. 9 compares the mathematical model proposed
by W;hitsitt er al. with the model described in this
paper, and relates the acid spending distance to the
acid effectiveness number (a term defined by Whitsitt as 0.92/Peclet number) for a given set of well
conditions. The agreement between these calculations
is good, indicating that the mathematical models are
conlparable.
Acid penetration distances predicted by Whitsitt
et al. do not agree with predictions made using data
presented in Fig. 2. Since the mathematical models
are comparable, the mixing coefficient data used by
Whitsitt er al. must be different from those presented
here. In Fig. 10 (Fig. 1 in Ref. 16), an acid penetration distance of 150 ft is predicted under conditions
for which our proposed model predicts 35 ft. The effective mixing coefficient would have to be 3 X 10-

MIXING

COEFFICIENT

Width
Rock Type

(in.)

RL
RL
RL
RL
RL
RL
RL
RL
RL
RL
RL
RL
RL
iii
RL
RL
RL
RL
RL
RL
RL
RL
RL
SL
SL
SL
SD
SD
SD
SD
SD

0.375
0.375
0.375
0.375
0.375
0.375
0.375
0.375
0.375
0.375
0.375
0.375
0.375
~.375
0.375
0.375
0.375
0.375
0.375
0.250
0.152
0.152
0.231
0.320
0.350
0.375
0.375
0.375
0.375
0.375
0.375

DATA

Acid
Concentration
(percent)

Nr/e*

N.,

15
15
15
15
15
15
15
15
15
15
15
15
15
15
15
15
15
28
28
15
15
15
15
15
15
15
15
15
15
15
15

290
348
362
406
507
2,268
4,237
5,228
906
5,204
7,080
10,240
12,288
12,288
14,208
19,048
9,472
2,970
4,566
4,300
5,824
3,795
7,829
411
5,621
8,544
1,360
3,248
3,900
3,652
Q.6M
-, ---

0.30
0
0.15
0.23
0.30
0
0
0
0
0
1.72
0
1.88
2.73
2.00
0
0.94
0
0
0
0
0
0
0.26
0
0
0
0
0
0
Q

D,

(sq

X 10
ft/min)

1.03
0.53
0.93
0.88
0.70
4.46
4.94
9.38
1.35
6.59
7.56
12.10
15.99
13.17
21.71
32.84
10.08
4.40
6.00
4.73
3.09
1.49
5.89
0.27
4.17
6.30
0.20
0.41
0.50
2.01
G nA
*.-r

walls, limestone.
walls, Iimeatona.
walls, dolomita.

857

sq ft/min to obtain the larger penetration distance,


or about one-sixth the value measured in our study.
Mixing coefficients as small as 3 X 10-6 sq ft/min
would have to be obtained at very low injection rates,
with the fracture model oriented so that acid flow is
either up or down a vertical fracture. Mixing data
obtained under such conditions would not represent
those in the fracture, and using them could lead to incorrect evaluation of acid treatment designs.

Nomenclature
acid concentration
initial acid concentration
fluid loss coefficient when fluid losses are
controlled by viscous resistances or
fluid compression, ft/~~
fluid loss coefficient when fluid losses are
controlled by fluid loss additives, ft/
~m
effective mass transfer rate for acid, sq
ft/min
effective mass transfer rate for acid when
siirface kinetics are iniiniteiy fast, sq
ft/min
total fracture height, ft
height of fracture that accepts fluid, ft
injection rate into the fracture, bbl/min
total injection rate per foot of formation
height, bbl/min/ft
injection rate during laboratory experiments, cc/see
formation permeability, md

c: :

Cc =

Cw =

De, D =
Dw =

kg =
h. =
i=
i/H =
i* =
kf =

15%

w
Cc

= .05

HCI
FT

= .007

Nh =
IVP,

N=, =
N ~,. =
N., =
t =
T =

UO=
Z=
v =

~ =
w =
~=
P=

References
1. McGuire,W. J. and Sikora, V. J.: The Effect of Verti-

cal Fractures on Well Productivity,

Tram.,

AIME

( 1960) 219, 401-403.

and Fast, C. R.: OptimumFluid Characteristicsfor Fracture Extension,Dri/1. and Prod. Prac.,

2. Howard, G. C.

5. Hendrickson,

A. R., Rosene, R. B. and Wieland, D. R.:


Acid ReactIon Parameters and Reservoir Characteristics
Used in the Design of Acid Treatment, paper presented
~;;$
137th ACS meeting, Cleveland, Ohio, April 5-14,

IN

T=75

La =

at
time ? (measured from wellbore), ft
acid penetration distance along one wing
of the fracture (measured from the
wellbore), ft
dimensionless acid penetration distance,
0.713 LaFhJi
Peclet number for mass transfer, w;/2D,
Reynolds number, for fluid loss, 2w;p/,u
Reynolds number for flow along the
fracture, 2WU0p/IL
Schmidt number for mass transfer, p/DP
time, minutes
temperature, F
flow velocity at the inlet to the fracture,
ft/min
average flow velocity along the fracture,
ft/min
velocity component normal to the centerline of the fracture at some time, t,
ft/min
average fluid loss velocity at some time,
t,ft/min
fracture width, in.
viscosity of fracturing fluid, cp
fluid density, lb,n/cu ft

API (1957).
3. Kiel, O. M.: A New Hydraulic Fracturing Process:
J. Pet. Tech. (Jan., 1970) 89-96.
4. Geertsma, J. and de Klerk, F.: A Rapid Method of
Predicting Width and Extent of Hydraulically Induced
Fractures, J. Pef. Tech. (Dec., 1969) 1571-1581.

hn = 100

L = length of one wing of the fracture

FT/ ~N
F

6, Knox, J. A., Pollock, R. W. and Beecroft, W. H.: The


Chemical Retardation of Acid and How It Can Be
Utilized, J. Cdrr. Pet. Tech. (Jan.-March, 1965).
7. Dill, W. R.: Reaction Times of Hydrochloric-Acetic
Acid Solution on Limestone; paper presented at 16th
~~;t~~;;t Regional ACS Meeting, Oklahoma City, Dec.
,.
8, van Poollen, H. K. and Jargon, J. R.: How Conditions
Affect Reaction Rate of Well-Treating Acids, Oil and
Gas J. (Oct. 21, 1968) 84-91.
9. van Poollen, H. K.: How Acids Behave in Solution,
Oi/ and Gas J. (Sept. 25, 1967) 100-102.
10. Hendrickson, A. R., Hurst, R. E. and Wleland, D. R.:
Engineered Guide for Planning Acidizing Treatments
Trans.,
Based on Specific Reservoir Characteristics;
AIME (1960) 219, 16-23.
11. Barron, A. N., Hendrickson? A. R. and Wieland, D. R.:
The Effect of Flow on Acid Reactivity in a Carbonate
Fracture, J. Pet. Tech. (April, 1962) 409-415.

FLUID V OLUME PUMPED =


40,000 GAL.
)

20
30
40
INJECTION RATE - BPM

Fig.10-Comparison

of acid penetration distance


predicted by the proposed model and by

the model of Whitsitt et al.


858

50

12. Smith, C. F., Crowe, C. W. and Wieland, D. R.: Fracture Acidizing in High Temperature Limestone, paper
SPE 3008 presented at SPE 45th Annual Fall Meeting,
Houston, Oct. 4-7, 1970.
13, Nierode, D. E. and Williams, B. B.: Characteristics of
Acid Reaction in Limestone Formations; SOC. Pet. J3rg.
J. (Dec., 1971) 406-418.
JOURNAL

OF PETROLEUM

TECHNOLOGY

Williams, B. B.: Fluid Loss from Hydraulically Induced Fractures; J. Pet. Tech. OUIY, 1970) 882-888.
15. Broaddus, G. C. and Knox, J. A.: Intluence of Acid
Type and Quantity in Limestone Etching, paper presented at Mid-Continent Meeting, API, Wichita, Kans.,
March 3 I-April 2, 1965.
16. Whitsitt, N. F., Barrington, L. J. and Hantsah, R. R.:
A New Approach to Deep Well Acid Stimulation Design, The Western Co. (June 10, 1970).
17. Harris, O. E., Hendrickson, A. R. and Coulter, A. W.:
High Concentration Hydrochloric Acid Aids Stimulation Results in Carbonate Formation, 3, Pet. Tech.
(Oct., 1966) 1291-1296.
18. WMiams, B. B., Gldley, J. L., Guin, J. A. and Schechter,
R. S.: Characterization of Liquid-Solid Reactions, Hydrochloric Acid-Calcium Carbonate Reaction, Ind. and
14.

Eng. Chem. Fund. (Nov., 1970) 9, 589.


19. Sinclair, A. R.: The Effects of Heat Transfer in Deep
Well Fracturing, J. Pet. Tech. (Dec., 1971) 1484-1492.

20. Graham, J. W., Kerver, J. K. and Morgan, F. A.:


Method of Acidizing and Introducing a Corrosion Inhibition into a Hydrocarbon
Producing Formation.
U. S. Patent 3,167,123 (Jan. 26, 1965).
21. Terrill, R. M.: Heat Transfer in Laminar Flow Between Parallel Porous Plates, Infl. 1. oj Hear and Ma.m
Transfer

( 1965)

8, 1491-1497.

APPENDIX
Experimental

Procedures

Experiments to determine values for the effective mixing coefficient for hydrochloric acid reaction were
conducted using the following procedure:
1. Cores were prepared to fit into the equipment
described in Ref. 13.
To prepare rough-waiied cores a 3-in.-diameter
original
Petroleum

manuscript
of paper SPE 3720 received in Society
Engineers
office Oct. 8, 1971, Revised manuscript

ceived March ~~, IaZZ. c COPYFirtht 1972 American


Mining,
Metallurgical.
and Petroleum
Engineers,
Inc.
This paper will be printed
covar 1972.

mm%-,

i972

in Transactions

volume

Institute

253, which

core was fractured in tension by applying a force to


opposite sides of the core. The core halves were then
fastened together and machined to fit into the equipment. To insure that the core could be sealed into the
equipment, the outside edge of the core was coated
with epoxy before the final machining. When the
cores were mounted in the equipment, the core sections fortned a fracture with very rough parallel walls.
Smooth-walled cores were cut from a block of reek
and machined to fit into the test equipment. To insure
a seal, outer core surfaces were coated with epoxy
before the final machining step.
2. Cores were mounted in the test equipment with
the channel between cores oriented to represent a
vertical fracture.
3. Flow velocity along the fracture and fluid loss
velocity were set, using water, and. an experiment was
begun by switching from water to acid.
4. Samples of acid effluent from the fracture were
collected and the extent of reaction was determined
by a standard EDTA titration. To achieve a measurable change in concentration when reaction rate was
low, the acid was cycled through the fracture several
times.
5. The effective mixing coefficients that would be
required to allow the mathematical model to fit experimental data were then calculated. Data were analyzed using a numerical solution of Eq. 1. Mixing
coefficients for experiments without fluid loss can be
evaluated using Eq. A-1.

of

re.
of
will

J-PI

Vous aimerez peut-être aussi