You are on page 1of 13

Theoretical and Experimental Studies

of Sandstone Acidizing
A.D. Hill, * SPE, Texas Petroleum Research Committee
D.M. Lindsay, Texas Petroleum Research Committee
I.H. Silberberg, SPE, Texas Petroleum Research Committee
R.S. Schechter, SPE, U. of Texas
Abstract
The matrix acidization of sandstone by a
hydrochloric/hydrofluoric acid mixture is described
through use of a capillary model. The model was
solved first in linear coordinates so that it could be
compared with the results of coreflood experiments
performed on Berea sandstone. The model
predictions showed reasonable agreement with the
experimental data and yielded specific information
about the reaction characteristics of the sand-
stone/HCIIHF system.
The acidization model then was applied in radial
coordinates to generate design curves for a matrix
acidization treatment. While these curves strictly
apply only to those sandstones having similar
mineral compositions, the approach is general. It is
based on matching the location of the HF reaction
front to the depth of a damaged zone. This method
introduces the concept of an optimum injection rate
and, in this regard, differs from other design
methods reported in the literature.
Introduction
The matrix acidization of sandstone by an HCIIHF
acid mixture is an often-employed oil well stimulation
technique designed to increase permeability in a zone
around the wellbore. The acid mixture flowing into
the porous medium reacts with the various mineral
species present, thus effecting an increase in the
matrix porosity and, it is hoped, the permeability.
Clearly, one of the factors controlling the depth of
acid penetration is the chemical composition of the
minerals which the acid contacts. Smith and Hen-
drickson,l Gatewood,
2
and Lund et al.
3
-
5
have
shown that the reaction with calcite is more rapid
'Now with Marathon Oil Co.
0197-7520/81/0002-6607$00.25
Copyright 1981 Society of Petroleum Engineers of AIME
30
than with silicate minerals (clay or feldspar), which
is, in turn, more rapid than the reaction with silica.
Several papers describing the distance of
penetration have been published. Smith and Hen-
drickson 1 and Smith et al.
6
first suggested the use of
linear core tests to predict radial penetration. Farley
et al.
7
reported tests similar to those conducted by
Smith and Hendrickson but measured many ad-
ditional parameters including the effluent acid
concentration, which is quite useful since the effluent
concentrations may yield information about reaction
characteristics. Experiments conducted in linear
systems are difficult to translate in terms of
penetration in a radial system, since the fluid velocity
varies inversely with radial distance. The obvious
approach has been to develop a mathematical model
that can be calibrated based on linear flow data and
then applied to a radial system.
Gatewood
2
proposed that the acid penetration
distance be determined by assuming that the reaction
of HF with the silicate minerals is much faster than
with the silica. The distance of penetration is
determined in this model by the formation com-
position and by the stoichiometry of the reactions.
Lund et al.
5
,8 and Fogler and McCune
9
developed a
model which neglects the reaction of HF with silica
but does consider the reactions with the silicate
minerals. The advantage of these approaches is that
the penetration depth can be predicted based on the
formation composition. However, the reaction with
silica cannot be neglected in determining the depth of
penetration, as will be seen.
Williams and Whiteley 10 used a somewhat dif-
ferent approach which includes an empirical
determination of the reaction rate based on linear
core flood experiments. The analysis assumes that a
quasistationary state exists. Williams 11 used these
empirical reaction rates to ascertain the total acid
SOCIETY OF PETROLEUM ENGINEERS JOURNAL
volume and injection rate needed to obtain a given
penetration distance. This approach includes the
silica reaction in the kinetic model, since the model is
determined empirically, and requires a separate test
at reservoir temperature of each new formation to be
acidized. Further, the analysis method that assumes a
quasistationary state is not valid for short times.
The conclusion provided by the analyses of both
Lund and Fogler
8
and Williams II is that the depth of
penetration increases with acid velocity. However,
some field experience indicates that injecting at the
maximum rate may not be the best strategy when the
volume of acid is limited.
12
This question is ad-
dressed in this paper.
Matrix acidizing is generally effective only when
applied to remove near-well bore damage. 13
Otherwise, the stimulation resulting from a treatment
will be quite small and is almost independent of the
permeability increase in the acidized zone. A design
to implement this strategy can be developed by
focusing primarily on the depth of acid penetration
and ensuring sufficient volumes to at least restore the
original permeabilities. This feature is a fortunate
one, because permeability changes caused by
acidization are complex, as evidenced by the data of
Smith and Hendrickson. I Their results indicate that
the permeability first declines and then increases.
There are two possible mechanisms for this
phenomenon, both of which probably contributed to
the observed permeability decline. Labrid 14
discussed the mechanism of permeability reduction
by precipitation of reaction products. However, the
movement of fines also may be responsible for the
observed permeability reduction. Evaluating the
relative importance of these two mechanisms is
difficult.
Acid Balances
Schechter and Gidley l5 and Guin et al.
16

17
have
developed a capillary model which predicts the
change in pore-size distribution resulting from acid
attack. The model approximates the porous medium
as a collection of cylindrical pores of varied sizes that
become enlarged as a result of the acid reaction at the
pore walls and allows for "collisions" between
pores. The pore structure is characterized by a pore-
size distribution 1'/ (A,x, t), where 1'/ (A,x, t) dA is
defined as the number of pores per unit volume
having a cross-sectional area between A and A +
dA. The change in 1'/ as a function of time is described
by an integrodifferential equation given in Appendix
A. Also shown in Appendix A are the equations
relating the permeability and porosity to moments of
the pore-size distribution.
The rate of acid reaction is an important feature
and is characterized as follows.
dA
- = 1/; (A,x,t). . .................... (1)
dt
The pore growth function 1/; depends on A because,
in general, acid must diffuse to the mineral surface to
react, and it depends on both time and position
because the local acid concentration must depend on
these factors.
FEBRUARY 1981
In sandstone acidization, many reactions take
place; the most important of them are the reactions
of HF and HCI with carbonates, the reaction of HF
with silicates such as clays and feldspars, and the
reaction of HF with quartz. To model the acid
concentrations as the acid mixture flows through and
reacts with the porous medium, acid balances for HF
and HCI must be written as follows.
a(C
HF
) aC
HF
+u--
at ax
-LXHF r
lO
1/;HF 1'/ (A,x,t)dA .......... (2)
o
and
a (C
HCl
) ac
HCl
+ u---
at ax
= - Lx HCl I
oo
1/;HCl 1'/ (A,x, t) dA , ........ (3)
o
where u is the flux, Xi is the moles of acid i expended
per volume of rock dissolved, and C
i
is the con-
centration of acid i.
The function 1/; is related directly to the overall
reaction rate as shown by Eq. 4.
2V7rRaA V2
1/; = , ...................... (4)
P
s
where ex is a stoichiometric coefficient, Ps is the
density of the solid, and R is the average overall
reaction rate over the entire reactive surface of one
pore. This overall rate depends on a series of in-
dividual processes including diffusion of the reac-
tants to the solid surfaces, reaction with the surface,
and diffusion of the products from the surface. In
extreme cases, the overall reaction rate may be
diffusion controlled or, in others, reaction limited.
Guin 16 has developed overall reaction rates for a
number of cases, and these general results need not
be repeated here. There are two limiting cases that
apply to this study. In pores of the size range
characteristic of sandstones, the reaction rate for the
dissolution of carbonates is much larger than the rate
of diffusion, and this permits use of the diffusion-
controlled approximations given in Appendix B.
The second limiting case that applies to the
reactions of HF with silicates (feldspars and clays)
and with silica also is given in Appendix B in a form
developed assuming that these reactions are first-
order in HF concentration. This approximation has
been shown to be a good one for the minerals of
interest in this work.
4

I1
The reaction rate constant of HF with quartz
differs from that with silicates so that the propor-
tions of these minerals present must be taken into
account. They may be combined to yield
-rHF = [ksilfclay + k
qtz
(1- fclay)]C
HF
, ... (5)
where fclay is the fraction of the surface area oc-
cupied by clay minerals.
In this kinetic model all the clays and feldspars
have been lumped into one average parameter. This
31
Mineral
Quartz
TABLE 1 - MINERAL CONTENT BY
PETROGRAPHIC ANALYSIS
Dolomite and siderite
Chlorite and illite
Feldspar
Percent of
Total Rock
75 5
10 3
10 3
53
combining is not necessary, but, on the basis of the
system studied, the approximation appears to be
valid.
A dynamic description of the parameter fcJay is
now necessary to track the changing average reaction
rate constant and permit the correct evaluation of the
lfHF function. The two rate laws for the individual
reactions of HF with silicates and quartz may be
written as
- rHF = kSi! C
HF
...................... (6)
and
-rHF = kqtzCHF ...................... (7)
Recalling that the reaction rates are expressed as
mass of acid reacted per unit area per unit time, then
dVsi!
dt
=
CisiI ksiI CHFSsiI
Psi!
.............. (8)
and
_ dvqtz = CiqtzkqtzCHFSqtz
, ............ (9)
dt Pqtz
where vsiI and V
qtz
are the volumes of silicate and
quartz per unit volume and SsiI and Sqtz are the
surface areas per unit volume of the two minerals,
respectively.
If it is assumed that the fraction of the exposed
surface area occupied by each mineral is propor-
tional to the volume per unit volume, there results
d (v tfcJay) Cisi! ksi!
- --fcJayStCHF ....... (10)
dt PsiI
and
d[v
t
(1- fcJay)]
dt
Pqtz
. (1- fcJay)StCHF , ................... (11)
where St = 2h L (" A Y2 1/ (A,x,t)dA.
o
Using the chain rule of differentiation and solving
each equation for dVtldt results in the following
equations.
dV
t
dt
_ dfcIay _ qsiI ksiI S C (12)
t HF
fcJay dt PsiI
and
dVt v t dfcJay _ Ciqtzkqtz S C
t HF .. (13)
dt 1 - fcJay dt Pqtz
Equating Eqs. 12 and 13 and recognizing that v
t
(total volume of solid per unit volume) is 1 - C/>, the
differential equation describingfcJay is obtained.
32
dfcIay = (Ciqtzkqtz _ CisiI ksiI )
dt Pqtz Psi!
. fcJay (1 - fcJay) S C
1- c/> t HF
.......... (14)
An examination of Eq. 14 reveals it to have the
correct qualitative behavior - i.e., for ksi! ap-
preciably greater than k
qtz
, fcJay will decline but at a
progressively slower rate as the silicate minerals
dissolve.
This equation, coupled with the acid balances,
predicts the acid concentrations throughout the
linear flow regime. It also is coupled with the
evolution equation, from which is obtained 1/ (the
pore-size distribution function) at all space positions
through time. A solution of this coupled system. of
integrodifferential, partial differential, and ordinary
differential equations will yield the desired predic-
tions of acid concentrations plus matrix porosity and
permeability values. The numerical techniques used
to solve this set of equations are described briefly in
Appendix C and in full detail elsewhere. 19
Similar approaches have been used by Lund et al.
5
and by Williams and Whiteley. 10 These studies both
employ empirically determined kinetic data to design
an acid treatment and do not consider the pore
geometry. Lund et al. also have neglected the
reaction with silica, whereas Williams and Whiteley
have analyzed their results assuming that a
quasistationary acid front is achieved. This latter
assumption is not valid during the initial stages of the
treatment, and neglecting the reaction of HF with
quartz is never valid, as is discussed in a subsequent
section.
The model presented here uses surface reaction
rates, which should apply independently of the
composition of the formation (only fcJay must be
adjusted to allow for compositional variations), and
does not make the simplifying assumptions imposed
by Lund et al. or Williams and Whiteley with regard
to reaction rates. In addition to considering the
reactions with HF, it was found necessary to account
for the reactions with HCI by including an HCI
balance. Previous studies have neglected this latter
reaction.
Experiments
To test the predictions of the capillary model, a series
of coreflood acidization experiments were per-
formed.
23
These experiments all were conducted
using Berea sandstone cores with initial composition
shown in Table 1 and an acid mixture of 3 wtOJo HF
and 12 wtOJo HCr.
The core (0.1 m long and 0.0254 m in diameter)
was placed in a Hassler holder, which then was
mounted into a nitrogen porosimeter, and the pore
volume was measured. Suitable corrections for the
void volume of the holder were obtained by
calibrating with a solid plug having the same external
dimensions as the Berea core.
After the initial porosity was determined, the
Hassler core holder (with core) was mounted into a
Ruska constant-temperature oven and allowed to
SOCIETY OF PETROLEUM ENGINEERS JOURNAL
reach thermal equilibrium at the desired reaction
temperature. The core was evacuated and filled with
a 3 wtOJo NaCl solution, and the initial permeability
was measured. The flow rate was determined in all
cases by employing a Ruska proportioning pump. A
backpressure regulator was used to maintain a system
pressure of 5 to 7 MPa. This pressure level is
necessary to prevent CO
2
bubbles from forming in
the core.
Each experiment consisted of the injection of at
least 90 PV of acid. The effluent was collected in
either 1- or 2-PV samples using a sample collector.
Pressure drops across the cores were measured
during the course of the experiment.
After the core holder had been removed from the
flow system, it was mounted into the porosimeter
apparatus without having to remove the core from
the holder. The final porosity was determined.
The experimental studies required a reliable
technique for determining the amount of
hydrofluoric acid that reacted. The procedure used
by Williams and Whiteley!! was adopted. The
method involves an ionometric technique which is
rapid and effective if carefully calibrated by
measurement of the amount of silica dissolved in the
effluent acid solutions. This two-step approach is
made necessary by the presence of fluosilicic acid and
other fluosilicates in the residual acid solutions. The
residual acid solution must be buffered before
analysis to a pH of 4.2 to 4.5 to ensure complete
dissociation of all hydrofluoric acid. At this pH level
the fluosilicates partially dissociate and release
fluoride ions. These then are included in the con-
centration sensed by the ion specific electrode. To
compensate, a glass slide etch procedure which
allows for direct determination of reactive HF was
used. The details of this procedure are available. 23
It was found that a new calibration curve is
required if either the proportions of HF and HCl or
the mineral content of the core are changed. If
sufficient care is exercised, the ionometric analysis is
rapid and reliable.
Discussion of Results
Effluent Concentrations
Typical acid (HF) effluent curves are shown in Figs. 1
through 4. Theoretical calculations are complex, and
the numerical techniques are described in Appendix
C. These results show that initial acid concentrations
are small but increase as more acid is injected until a
plateau value less than the injected acid con-
centration is reached. Qualitatively, this profile can
be understood in terms of the differences in the
reaction rates between the HF and the minerals
present. The carbonates are expected to react first,
with the reactions with silicates and quartz following
in that order. The importance of the HF/quartz
reaction is shown by the plateau region that develops
in the latter stages of the experiment. This plateau
provides a measure of the HF/quartz reaction, as
most of the other minerals that are accessible should
have dissolved by this stage of the experiment. The
level of this plateau indicates that the HF/quartz
FEBRUARY 1981
T= 25.0C a", 0.342 MLlSEe
EXPER\MENTRL DATA. RUN NO. -4
- MODEl PRED1CllDN
"t.aa 20.00 30.00 40.00 50.00 60.00 70.00 80.00
PORE VOLUMES OF RCID INJECTED
Fig. 1 - Comparison of predicted and experimental ef-
fluent acid concentrations (Run 4) for initial
concentrations of 3 wt% HF and 12 wt% HC!.
T", 2S.0C 0= 0.250 tiLiSEC
(!) E'(PER1MENTAL OqHL RUN NO.5
- MODEL PREGICT ION
2000 3000 40.00 50.00 50.00 CjO.oo
vOLUMES OF ACID iNJECTED
Fig. 2 - Comparison of predicted and experimental ef-
fluent acid concentrations (Run 5) for initial
concentrations of 3 wt% HF and 12 wt% HC!.

=>

0
T" 2S.0C Q", 0.130 r)L/5EC
(!) [XPERIMENHll DATA. RUN NO.6
- MODEL PREDICT JON
10.00 20.0:) J:J.O!l (:J.OO so.::m 60.00 70.00 "10.00
PORf "OuJl1ES ACID iNJECTED
Fig. 3 - Comparison of predicted and experimental ef-
fluent acid concentrations (Run 6) for initial
concentrations of 3 wt% HF and 12 wt% HC!.
33
q T"" 2S .OC Q= 0.063 MLlSEe
(!) EXPER I MENTAL DATA. RUN NO. 7
- MODEL PREDICT ION
. . .
.....
g
<\,:+,.,,,.-,--,,:,,-".,,:===::;,,:--:.,::-, ----=,""",.,::-,---,."-,.=,,-s"',.-:c,,c---:<.'-:c."=--"70,-;;.';O-, ----;1,,:--:.,::-,----;;)90.00
PORE VOLUMES OF ACID INJECTED
Fig. 4 - Comparison of predicted and experimental ef
fluent acid concentrations (Run 7) for initial
concentrations of 3 wt% HF and 12 wt% HCI.
u

T= 25 .OC ao: 0.250 MLISEC
(!) EXPER I MENTAL DATA. RUN NO. 5
{!) FRACTION OF CLAY=.D5
X FRACTION OF CLAY=.166
.

9J.00 10.00 20.00 30.00 .. 0.00 50.00 60.00 10.00 80.00 !fO.OO
PORE VOLUMES OF RCID INJECTED
Fig. 5 - Effect of parameter Ac1ay(O) on effluent acid con
centration predicted by the model.
52.0C Q= 0.128 t'lL'Sf[
(') DqTR. RuN NO.9
- MODfl ION
"'0


. .
0 0
'"
0

......... ;,,,-"'. ,,=-';:zo-:. ,,:--",,:-:. ,-=-,--:1,,:--:. ,=-, ----",,:--:. ,::-, ----:,,". ,::-, ----:,';-,. ,::-,----:,"-, .=" ---;:90.00
R[ID INJECTED
Fig. 6 - Comparison of predicted and experimental ef
fluent acid concentrations (Run 8) for initial
concentrations of 3 wt% HF and 12 wt% HCI.
34
reaction is significant and cannot be neglected if the
depth of acid penetration is a concern.
To clarify this point further, it may be noted that
at 52e the stabilized acid effluent concentration
ratio is about 0.6 as shown in Figs. 6 and 7. Thus,
40070 of the HF is consumed within 0.1 m, the length
of the core. Thus, even though the reaction of HF
with quartz is relatively slow, it is still an important
feature of sandstone acidizing and cannot be
neglected.
By using the first -order reaction rate constant
reported for feldspar
4
as ksil and by empirically
adjusting k tz and fela (0) (the fraction of the
reactive surface occupied by the silicates after the
carbonates have been removed), the effluent com-
positions obtained by integrating the acid balance
equations could be brought into reasonable,. but
certainly not perfect, agreement with the ex-
perimental data as shown in Figs. 1 through 4. The
values of the parameters found to give reasonable
representation of the data at 25e were as follows.
= 7.6 x 10 - 6 __ k_g_H_F __ _
2 (kmOI HF)
m s
m
3
(from Fogler
4
) .
kg HF
k
qtz
=5.0x 10-
8
-------
2 (kmOI HF)
m 's
m
3
felay(O) = 0.05 .
The empirically determined k
qtz
at first appeared to
be too small when compareo with the results of
Blumberg,18 Born,24 Mowrey,25 and Guin.
16
These
investigators all report reaction rate constants of
apf,roximately 5 x 10-
7
kg HF/m
2
s (kmol HFI
m ) for the reaction of HF with quartz at 25e.
However, those investigations all were conducted
using amorphous silica, whereas most if not all of the
silica found in sandstones is a-quartz. It is well
known that the solubility of amorphous silica in
water is generally (dependinf. on the temperature)
larger than that of a-quartz. 2 ,26 Only one paper was
found reporting the reaction of a-quartz with HF at
the concentration levels of interest in this study.
Analysis of the data presented by Bergman
28
yielded
a first-order rate constant at 25e between
4.2 x 10-
8
and 6.8 x 10-
8
kg HF/m s (kmol
HF/m
3
). These values compare favorably with the
rate constant which was found to fit the experimental
data and are consistent with the reported differences
in solubility of a-quartz as compared with amor-
phous silica.
The second parameter that was adjusted to obtain
a reasonable fit of the effluent acid concentration
profiles was felay(O), the fraction of the reactive
surface occupied by silicates after the carbonates
have been reacted. Assuming fclay(O) to be equal to
the volume fraction of silicates present (0.16) gave
unrealistic predictions of the acid effluent con-
centration; a value of 0.05 yielded a much better fit
of the data, as shown in Fig. 5. Thus, it appears that
SOCIETY OF PETROLEUM ENGINEERS JOURNAL
all the silicates did not react at the high rate found for
feldspars.
To shed light on the significance of the parameter
fclay(O), a petrographic analysis was made of a
sandstone core that had been acidized with 90 PV of
acid mixture. This analysis consisted of examining a
thin section of the acidized rock with a petrographic
microscope. The study indicated that most of the
feldspars had been removed by the acid treatment,
whereas the clays present appeared to have reacted
only slightly. Thus, it appears that the parameter
fclay(O) is related to the fraction of feldspar present
initially (see Table 1). Furthermore, it appears for the
sandstone studied that most of the clay reacted at a
rate close to that of quartz.
An analysis of total aluminum present before and
after acidization showed that, in the front of the
acidized core, approximately 75OJo of the original
aluminum present had been removed. This result
indicates that the clay minerals reacted with the acid
mixture to some extent and that the aluminum layers
in the clays possibly were leached preferentially. Such
a preferential attack of acid on the octahedral layers
of a clay has been found previously for the reaction
ofHCI with clay.29
Note that the clay minerals in the Berea sandstone
were in the form of rock fragments such as pieces of
mica and shale. The reactivity to clay in this state
cannot be generalized to other types of clay com-
monly found in sandstones. In particular, the specific
surface area of clay in this state may be much less
than that found for clay in the form of small
platelets, thus greatly reducing the reactivity.
Using a similar fitting procedure and fclay(O) =
0.05, values for k
qtz
and ksil were obtained that gave
reasonable fits ot the experimental data at 52C
(Figs. 6 and 7). The values used at this temperature
were
and
Permeability Response
The matrix permeabilities during acidization
predicted by the model did not agree well with those
found experimentally (Fig. 8). As the capillary model
used in this study had given excellent predictions of
permeability changes in homogeneous matrix
materials in previous studies, 16,30 it is presumed that
the failure of the model when applied to sandstone is
due primarily to the heterogeneity of the material.
Experimental permeability profiles during
acidization usually show an initial decline in per-
meability, due most likely to the downstream
migration of fine particles as cementing substances
are dissolved. Such a permeability decline cannot be
predicted by this capillary model, since particle
migration is not accounted for. A further difficulty
FEBRUARY 1981
1:: 52. DC Q", 0.250 MLiSEC
(!) EXPERIMENTAL DATA. RUN NO.9
- MODEL PREDICTION
10.00 20.00 JO.oo '50.00 6::1.00 10.00 80.00 90.0?
PORE VOLUMES OF RC! 0 ! N JEUEO
Fig. 7 - Comparison of predicted and experimental ef
fluent acid concentrations (Run 9) for initial
concentrations of 3 wt% HF and 12 wt% HCI.
o

I I I
7 -
6
5
2 f-
1 ,..
Acid Rate: O. 250 ml/sec
Temperature: 23.5C
CD Experimental Data, Run No.5
- Model Prediction
-

-
-
-
-
-
-
-
____ ____ ____ ____ i ____ --J
o 20 40 60 80 100
PORE VOLUMES OF ACID INJECTED
Fig. 8 - Comparison of permeability prediction with ex
perimental results for initial concentrations of 3
wt% HF and 12wt% HCI.
35
1.5
4>0
0.15
0
-e-
Vr
"-
0.153
-e-

-
0

I- 1200
ct r.
a:
1.25
>-
q
1.15
t:
qo
(J)
0
a:
0
Q.
1.0
1.0
Fig. 9 - The optimal strategy for an undamaged formation.
encountered in predicting permeability changes arose
from the necessity to use the approximate growth
function shown in Appendix B to model the dif-
fusion-controlled reaction of HF and HCI with
carbonates, because the exact function created a
numerical instability in the evolution equation. !6
Design of an Acid Treatment
Optimal Strategy
In the final analysis, economic considerations govern
the design of an acid treatment. However, even given
the same technical information, the "best" treatment
may vary from company to company because
economic criteria vary. Instead of determining that
most profitable action, it is possible to calculate the
greatest stimulation that can be produced using a
given volume of acid. Given the solution to this
problem, which is called the optimal strategy, then
the best course of action defined in some economic
sense can be selected.
To obtain the optimal strategy, it is assumed that
the acid reaction rate is so well controlled that it is
possible to increase the porosity in the formation
surrounding the well bore in any desirable way. Thus,
4>(r) (the porosity following treatment) is the control
variable. The volume of rock dissolved by an acid
treatment is
rd
Vr = 27rh r(4)-4>o)dr . ............... (15)
Here, 4>0 is the porosity before treatment, h is the
thickness of the formation to be treated, and r wand
r d are the radii of the well bore and drainage area,
respectively. The volume of rock dissolved by a
volume of acid is fixed by stoichiometry; thus,
specifying the treatment volume fixes Vr- Using
Darcy's law for radial flow, the new production
resulting from the acid treatment is given by
................... (16)
36
Zone Initially
4>0
0.15
Damaged
3.0 I
Vr
0.153
0

-e-
I "-
-e-
I


rw
1200
0 I

k DAMAGED
0.01
a:
2.0 k FORMATION
>-
!:::
(J)
q
7.5
0
a:
1.5
qo
0
Q.
1.0
1.0 3.0 4.0
r / r.
Fig. 10 - The optimal strategy for a damaged
Here, ko (r) is the permeability before treatment,
while k(r) is the resulting one. The problem is to
maximize qlqo for a fixed V
r
. Since the numerator is
determined by the initial state of the reservoir, the
problem reduces to
min [(d dr ]
(r) J rk(r) , ..................... (17)
rw
subject to Vr being constant. To proceed, the per-
meability must be related to the change in porosity.
Lund and Fogler
8
have proposed
= f .................... (18)
This expression is used here, although more complex
relationships which better define the response of a
given formation may be substituted easily.)!
Pontryagin's maximum principle defines the
solution of variational problems such as that posed
here. Using this principle, it can be shown that
4> [ n ]lI(n+!)
= 2 for 4> ;::: 4>0
4>0 {3r 4>oko (r)
............................. (19)
Given the volume of acid to be used, the constant (3
can be evaluated by substituting Eq. 19 into Eq. 15
and 4>(r) then determined. The function 4>(r)
represents the new porosity at each point r which the
acid is to create if the maximum stimulation is to be
obtained. Obviously, it is not possible to tailor the
acid reaction rate so that this optimal porosity
distribution will result; however, the desired porosity
distribution should be approached as nearly as
possible.
The optimal porosity distributions for two dif-
ferent cases are shown in Figs. 9 and 10. For an
undamaged reservoir, the maximum stimulation
ratio is 1.15 when about 1.8 m
3
of acid per meter of
formation thickness of 3.0 wtOJo HF and 12 wt% HCI
mixture are used in a 0.3-m-diameter wellbore. Large
stimulation ratios can be achieved if there is near-
well bore damage. The optimal porosity distribution
for the case in which the damaged zone extends a
SOCIETY OF PETROLEUM ENGINEERS JOURNAL
distance 1.5 r w is shown in Fig. 10. Note the
discontinuity in the porosity that occurs at the outer
limit of the damaged zone, indicating that the
maximum stimulation is achieved when the greatest
portion of the acid is expended in removing damage.
The stimulation ratio for this case is 7.5, indicating
that large increases in production can be achieved by
damage removal.
These calculations provide a measure of the best
that can be achieved given a carefully defined
situation. However, they do not provide a practical
means of designing an acid treatment.
Application of the Model
A sandstone acidization model will be useful if it can
yield information needed for the design of an actual
acidization treatment. Such information may be
obtained from the model used in this study by ap-
plying it in radial coordinates. It was found from the
model (expressed in radial coordinates) that a
quasistationary HF concentration profile is
established after the carbonate material has reacted
and that the HF front moves slowly, as can be seen in
Fig. 11. Furthermore, the position of the HF front
depends on the injection rate - the higher the flow
rate, the farther into the matrix the front will extend.
Williams 11 constructed acidization design curves
based on a definition of radius of permeability in-
crease as that radius at which the formation porosity
had reached a certain level. For a given desired radius
of permeability increase, such a definition
automatically requires that operation at the highest
injection rate possible will use the least total volume
of acid and, thus, be the most desirable treatment
scheme. Since the optimal strategy developed in the
previous section revealed that the most effective use
of a given volume of acid is to penetrate a damaged
zone, the Williams criterion, which requires the
highest injection rate, may not always be the
preferred one. Since the depth of acid penetration is a
sensitive function of the injection rate (a
quasistationary front is established), an alternative
design strategy is proposed as follows.
1. Choose an injection rate such that the HF acid
front reaches the desired radius of permeability
improvement. This radius should correspond to the
radius of damage.
'2. Inject enough total volume of acid at the chosen
rate so that the total amount of solid reacted in the
penetrated region reaches a certain level. This
amount is chosen from experimental permeability
date to ensure a permeability increase throughout the
region.
Based on this design strategy, illustrative design
curves have been generated by the model for sand-
stone with three different feldspar contents and a
well radius of 3 in. These curves are presented in
Figs. 12 through 14. The reaction rate constants at
the various temperatures needed to generate these
design curves were obtained from Arrhenius
relationships based on the rate constants known at 25
and 52C. Such an extrapolation of reaction rates to
higher temperatures is not always valid but was
FEBRUARY 1981


C)

i:i o.

S o.
5
C)
0

.,
.. o.
i!l
:l
Injection rate: 0.1 bbl/min/n of reservoir thickness
Temperature: 200F
v = 66 gal/It of reservoir thickness
S
RADIAL mSTANCE FROM WELLBORE (INCHES)
Fig. 11 - HF concentration profiles in a radial flow system
for initial concentrations of 3 wt% HF and 12
wt% HCI.
w
E-o
0(
p::

E=:
t.J
w
....,

0.7
0.6

w

t.J 0.5
53
E-o
p::
(5
> 0.4
p::
w
00
w
p::
0.3
0
E-o

.......

;:;:: 0.2
.......
....l
ill

0.1
0.0 t-......; .... --"---........ ---.... ---I 2500F
200F
150F
lOOF
____ ____ ______ ____
o 3 6 9 12
PENETRATION DISTANCE (INCHES)
Fig. 12 - Design curves for 3 wt% HF and 12 wt% HCI in
2%-feldspar formation.
37
necessitated by the lack of reaction rate information
above 52e.
An inspection of Figs. 12 through 14 shows that, to
a reasonable approximation, the total acid volume
required to increase substantially the permeability
throughout a region about the wellbore is sensitive to
the formation composition but not to temperature.
Thus, the recommended acid volume (3 wtOJo HF and
12 wt% HCl) depends on the depth of the damaged
zone and the formation composition. The optimal
injection rate depends on all of the variables but is
most sensitive to temperature. At temperatures in
excess of 93.3C, it is not possible to obtain deep
penetration because the injection rates required
cannot be achieved in most cases without fracturing
the formation.
The most important conclusion is that the highest

o. 1
o. 0
300
200
100
o
o 3 8 9 12
PENETRATION DISTANCE (INCHES)
Fig. 13 - Design curves for 3 wt% HF and 12 wt% HCI in
5%feldspar formation.
38
rate is not always the optimum. Damage near the
well bore may be removed best by relatively small
treatments applied at modest rates.
The results can be viewed in another way. Once the
volume of acid is selected, the best rate then is given.
The depth of acid penetration also is fixed. Ob-
viously, these design curves can be used best when the
extent of the damaged zone is known; however, their
use also is recommended if the acid volume is
prescribed, perhaps on the basis of experience in a
particular formation.
In extending this design procedure to types of
sandstones other than Berea, it is anticipated that the
reaction rates of some clay minerals will be quite
different from those found in Berea. Unfortunately,
data on reaction kinetics are not now available. It
will be necessary to establish these reaction rates.and
00
UJ
r.:l

0.6

u 0.5
53
E-<
0.4
o. 1
______ ______ ______ ______
300
200
100
OL-__ ___ """' ___ __ --J
o 3 6 9 12
PENETRATION DISTANCE (INCHES)
Fig. 14 - Design curves for 3 wt% HF and 12 wt% HCI in
10%feldspar formation.
SOCIETY OF PETROLEUM ENGINEERS JOURNAL
their temperature dependence to develop design
curves similar to those presented here for Berea.
A second possible difficulty is that often the
damage is a result of drilling fluid invasion. The clays
present in the fluid may react at a different rate,
thereby altering the reaction rate. In turn, both the
penetration distance and volume of acid needed to
achieve good stimulation will be altered.
Conclusions
Matrix acidizing often is conducted at the highest
rate the formation can accept without fracturing. It
has been argued correctly that increased rates will
result in increased depths of acid penetration. The
strategy used here differs from this one. Given a
limited acid volume, it was found best to adjust the
rate so as to increase effectively the permeability of a
near-well bore damage zone. To implement this
strategy, a means of predicting acid penetration
depth is needed. A model which uses both the
chemical composition of the formation and its pore-
size distribution has been derived. Although the
model did not represent accurately the observed
variations in permeability, it did predict measured
acid effluent concentrations.
In addition to the development of a design strategy
and illustrating its implementation, the following
conclusions have been drawn from this study.
1. The capillary model predicts effluent acid
concentrations well.
2. Rate constants determined independently were
used for the reactions of HF with quartz and with
silicates, and the model yielded reasonable fits to the
experimental data. This agreement provided support
for the validity of the model.
3. The feldspar in Berea sandstone reacts at a
higher rate than the clays, which are present
primarily in rock fragments. The low reactivity of
clay observed in this study could be due to the fact
that the surface area available for attack was much
smaller in those rock fragments than that found for
clays with a fine platelet surface.
4. Though the clays in this study appeared to react
at a slow rate similar to that of quartz, it is not
known at what rate the fine clay particles, sometimes
responsible for well damage, will react. The reaction
rates of such materials would affect the design
results.
5. Illustrative design curves were obtained from the
model in radial geometry based on the location of the
HF front. This criterion shows that the acid injection
rate should be chosen based on the desired radius of
permeability increase and that neither the fastest nor
the slowest rate is always optimal.
6. The reaction of quartz with HF is not negligible
as has been assumed in other works
9
but is, in fact,
crucial to the effectiveness of sandstone acidization.
Nomenclature
A = cross-sectional area of a pore, m
2
(sq ft)
A p = reference pore cross-sectional area, m
2
(sq ft)
FEBRUARY 1981
C
j
= concentration of acid i, mol/m
3
(mol/cu ft)
D j = molecular diffusivity of acid i, m
2
Is
(sq ft/s)
fcJay
fraction of surface area occupied by
reactive silicates after dissolution of
carbonates
F G = geometric factor
h = formation thickness, m (ft)
k = permeability, m
2
(md)
k(r) = permeability after acid treatment at
radius r, m
2
(md)
ko (r) = permeability before acid treatment at
radius r, m
2
(md)
k
j
= first-order reaction rate constant,
kg i/m2 s(mol HF/m
3
)
[Ibm ilsq fts(mol HF/cu ft)]
first-order reaction rate constant for
reaction of HF with quartz,
kg HF/m2 s(mol HF/m
3
)
[Ibm HF/sq fts(mol HF/cu ft)]
first-order reaction rate constant for
reaction of HF with silicates,
kg HF/m2 s(mol HF/m
3
)
[Ibm HF/sq fts(mol HF/cu ft)]
L = average pore length, m (ft)
M
j
= the jth moment of the pore-size
distribution function
n = exponent of porosity Ipermeability
relationship
q = well production after acid treatment,
m
3
Is (cu ft/s)
qo = well production prior to acid treatment,
m Is (cu ft/s)
r = radial space dimension, m (ft)
r d = drainage radius, m (ft)
- rHF rate of disappearance of HF,
kg HF/m . s (Ibm HF/sq fts)
-rHF
-rHF
rate of disappearance of HF due to
reaction of HF with silicates,
kg HF/m
2
s (Ibm HF/sq fts)
rate of disappearance of HF due to
reaction of HF with quartz,
kg HF/m
2
s (Ibm HF/sq fts)
r w = well radius, m (ft)
R average reaction in a pore,
kg acid/m
2
s (Ibm acid/sq fts)
specific surface area of quartz, m
2
/m
3
(sq ft/cu ft)
specific surface area of silicates, m
2
/m
3
(sq ftlcu ft)
specific surface area of all solids, m
2
/m
3
(sq ftlcu ft)
t = time, seconds
u = volumetric flux, m
3
1m2 s(cu ftlsq fts)
U
qtz
specific volume of quartz, m
3
/m
3
(cu ftlcu ft)
usil specific volume of silicates, m
3
1m
3
(cu ftlcu ft)
39
V
t
specific volume of all solids, m
3
/m
3
(cu ft/cu ft)
Vr = volume of rock dissolved, m
3
(cu ft)
x = space dimension in linear flow system, m
(ft)
Xi moles of acid expended per cubic meter
(35.3 cu ft) of rock dissolved
a stoichiometric factor (solid dissolved per
acid reacted), kg/kg (Ibm/Ibm)
{3 constant
r gamma function
1J pore-size distribution, m - 5 (ft - 5)
Ps density of a solid, kg/m
3
(Ibm/cu ft)
7 = transformed time variable
(r) porosity at radius r
c = porosity after carbonates have been
removed
o initial porosity
1/;_ pore growth function, m
2
/s (sq ft/s)
Acknowledgments
We thank the Texas Petroleum Research Committee
for sponsoring this work and the U. of Texas for
University Fellowship support of one of the authors.
We also express our appreciation to Halliburton
Services for conducting additional experiments and
to Charles R. Williamson for performing the
petrographic analyses.
References
1. Smith, C.F. and Hendrickson, A.R.: "Hydrofluoric Acid
Stimulation of Sandstone Reservoirs," J. Pet. Tech. (Feb.
1965) 215-222; Trans., AIME, 234.
2. Gatewood, J.R., Hall, B.E., Roberts, L.D., and Lasater,
R.M.: "Predicting Results of Sandstone Acidization," J. Pet.
Tech. (June 1970) 28, 693-700.
3. Lund, K., Fogler, H.S., and McCune, C.C.: "Acidization I -
On the Dissolution of Dolomite in HCI," Chem. Eng. Sci.
(March 1973) 28, 691-700.
4. Fogler, H.S. and Lund, K.: "Acidization III - The Kinetics
of the Dissolution of Sodium and Potassium Feldspars in
HF/HCI Mixtures," Chem. Eng. Sci. (Nov. 1975) 1325-1332.
5. Lund, K., Fogler, H.S., and McCune, C.C.: "Predicting the
Flow and Reaction of HCIIHF Acid Mixtures in Porous
Sandstone Cores," Soc. Pet. Eng. J. (Oct. 1976) 248-260;
Trans., AIME, 261.
6. Smith, C.F., Rose, W.M., and Hendrickson, A.R.:
"Hydrofluoric Acid Stimulation - Developments for Field
Application," paper SPE 1284 presented at SPE 40th Annual
Meeting, Denver, Oct. 3-6,1965.
7. Farley, J.T., Miller, B.M., and Schoettle, V.: "Design Criteria
for Matrix Stimulation with Hydrochloric-Hydrofluoric
Acid," J. Pet. Tech. (April 1970) 433-440.
8. Lund, K. and Fogler, H.S.: "Acidization V: The Predictions
of the Movement of Acid and Permeability Fronts in Sand-
stones," Chem. Eng. Sci. (May 1976) 31, 381-392.
9. Fogler, H.S. and McCune, C.C.: "On the Extension of the
Model of Matrix Acid Stimulation to Different Sandstones,"
A IChE J. (July 1976) 22, 799-805.
10. Williams, B.B. and Whiteley, M.E.: "Hydrofluoric Acid
Reaction with a Porous Sandstone," Soc. Pet. Eng. J. (Sept.
1971) 306-314; Trans., AIME, 251.
11. Williams, B.B.: "Hydrofluoric Acid Reaction with Sandstone
Formations," J. Eng. Ind., Trans., ASME (Feb. 1975) 252-
258.
12. Templeton, C.c., Richardson, E.A., Karnes, G.T., and
Lybarger, J.H.: "Self-Generating Mud Acid," J. Pet. Tech.
(Oct. 1975) 1199-1203.
40
13. Williams, B.B., Gidley, G.L., and Schechter, R.S.: Acidizing
Fundamentals, Monograph Series, Society of Petroleum
Engineers, Dallas (1979) 6, Chap. 2.
14. Labrid, J .C.: "Thermodynamic and Kinetic Aspects of
Argillaceous Sandstone Acidizing," Soc. Pet. Eng. J. (April
1975) 117-128.
15. Schechter, R.S. and Gidley, J.L.: "The Change in Pore Size
Distributions from Surface Reactions in Porous Media,"
AIChE J. (May 1969) 339-350.
16. Guin, J .A.: "Chemically Induced Changes in Porous Media,"
Report No. UT 69-2, Texas Petroleum Research Committee,
Austin, TX (Nov. 1969); PhD dissertation, U. of Texas,
Austin (1969).
17. Glover, M.C. and Guin, J.A.: "Dissolution of a
Homogeneous Porous Medium by Surface Reaction," AIChE
J. (Nov. 1973) 1190-1195.
18. Blumberg, A.A.: "Differential Thermal Analysis and
Heterogeneous Kinetics: The Reaction of Vitreous Silica with
Hydrofluoric Acid," J. Chem. Phys. (July 1959) 1129-1132.
19. Hill, A.D.: "Flow and Simultaneous Heterogeneous Reactions
in Porous Media," Report No. UT 78-1, Texas Petroleum
Research Committee, Austin, TX (1978); PhD disseration, U.
of Texas, Austin (1978).
20. Reigle, E.G.: "A Study of the Effect of Core Size on Apparent
Pore Size Distribution," MS thesis, U. of Texas, Austin (June
1962).
21. Smith, C.S.: "A Systems Engineering Approach to the
Simulation of Distributed Parameter Processes," PhD
dissertation, Heriot-Watt U., Edinburgh, Scotland (Oct.
1975).
22. Gear, C.W.: "The Automatic Integration of Ordinary Dif-
ferential Equations," Comm. ACM(March 1971) 185-190.
23. Lindsay, D.M.: "An Experimental Study of Sandstone
Acidization," Report No. UT 76-1, Texas Petroleum Research
Committee, Austin (July 1976).
24. Born, H.K.H.: "The Mechanism of the Dissolution of Silica in
Hydrochloric-Hydrofluoric Acid Mixtures," PhD disser-
tation, U. of Texas, Austin (May 1976).
25. Mowrey, S.L.: "The Theory of Matrix Acidization and the
Kinetics of Quartz-Hydrogen Fluoride Acid Reactions,"
Report No. UT 73-4, Texas Petroleum Research Committee,
Austin (Sept. 1973).
26. Wollast, R.: "The Silica Problem," Sea, M.N. Hill et al.
(eds.), John Wiley and Sons Inc., New York City (1974) 359-
392.
27. Owen, L.B.: "Precipitation of Amorphous Silica from High-
Temperature Hypersaline Geothermal Brine," Publication
OCRL-51866, Lawrence Livermore Nat!. Laboratory,
Livermore, CA (June 1975).
28. Bergman, 1.: "Silica Powders of Respirable Sizes IV. The
Long-Term Dissolution of Silica Powders in Dilute
Hydrofluoric Acid: An Anisotropic Mechanism of Dissolution
for the Coarser Quartz Powders," J. Appl. Chem. (Aug. 1963)
356-361.
29. Osthaus, B.B.: "Chemical Determination of Tetrahedral Ions
in Nontronite and Montmorillonite," Clays and Clay
Minerals, A. Swineford and N. Plummer (eds.) Nat!. Research
Council Publication 327, Washington, DC (1954).
30. Sinex, W.E. Jr., Schechter, R.S., and Silberberg, LH.:
"Dissolution of a Porous Matrix by Slowly Reacting Flowing
Acids," Ind. Eng. Chem. Fundam. (May 1972) 205-209.
31. Champion, L.S.: "The Ultimate Yield from Oil Well
Stimulation with Acids," MS thesis, U. of Texas, Austin
(1970).
APPENDIX A
Capillary Model
In this model, the porous medium is approximated as
a collection of cylindrical pores all having the same
length L.15-17 The distribution of pore areas is
defined by a pore-size distribution 1J (A,x, t), where
1J(A,x,t)dA is the number of pores per unit volume
having a cross-sectional area between A and A +
dA. It can be shown 15 that the two important matrix
SOCIETY OF PETROLEUM ENGINEERS JOURNAL
properties - porosity and permeability - are given
by
cf>(x,t) = L ~ A 1](A,x,t)dA ........... (A-I)
and
k(x,t)
o
ko r'A
2
1](A,X,t)dA
o
~ O O A 21] (A,x,O)dA
o
, ....... (A-2)
where ko is the initial permeability. It is evident that
knowledge of the pore-size distribution during the
acidization process would yield predictions of
porosity and permeability throughout the reaction
region. An equation, called the evolution equation,
describing the behavior of the pore-size distribution
has been derived based on the concepts that the pores
continuously are being enlarged due to acid attack
and that, as the pores become larger, the walls
separating adjacent pores may dissolve, leaving a
single pore having an area equal to the sum of the
original areas. These concepts can be expressed in
mathematical terms as
a1] + --!- ( 1/;1]) = L [( 1/; (A,X, t) 1] (A - A,X, t)
at vA 0
'1] (A,X,t) dA - ~ ext1/; (A,X, t)
o
+1/;(A,x,t) ]1](A,x,t)1](A,x,t)dA } ..... (A-3)
1/;(A,x,t) is the pore-growth function defined by
dA
dt = 1/;(A,x,t). . .................. (A-4)
Note that 1/; depends on the many factors governing
the overall reaction rate in a pore. These factors
include acid concentration, pore geometry, dif-
fusion, and surface reaction rate. Knowledge of what
takes place in a single pore is required to determine
the evolution of the pore sizes. However, the solution
of the integrodifferential equation (Eq. A-4) is
complex, and a numerical approach generally is
required.
APPENDIXB
Overall Reaction Rates
Guin 16 has developed overall reaction rates for a
number of cases, and these general results are not
discussed here. There are, however, two limiting
cases that apply to the present study. If the reaction
rate is much larger than the rate of diffusion, which
is the case for the dissolution of carbonates, the
overall rate is diffusion controlled and
u Ca
.f. - _1_ A2 for A < A (B la)
'f'HCI - L2M2 P
s
P , -
and
~ (36)2/3 aCi ~ ) 1/3 (D(7rA ) 213
1/;HCI r(1I3)p
s
M2 2L
for A > Ap , ...................... (B-Ib)
FEBRUARY 1981
where
-2
(
18.IDuiL M2) V2
Ap =
and Di is the molecular diffusivity of the acid. The
second limiting case is surface reaction controlled,
and
_ 2hakC
i
V2
1/;(A,t) - A , ............... (B-2)
Ps
where k is a surface-reaction rate constant.
APPENDIXC
Numerical Techniques
As long as the 1/; function is a separable function of
C
i
and A having the form
1/;i = fe(Ci)fA (A) , ................. (C-l)
the evolution equation may be transformed so that it
may be solved independently from the acid balance
equations.
a1] a[fA (A)1]] _ -[ roo (A-'" )
-+ --LJ1] I'\,T
aT aA 0
'1] (A,T)f
A
(A)dA- ~ o o VA (A) +fA (A)]
o
'1](A,T)17(A,T)dA} , ................. (C-2)
where
t
T(X,t) = fe(Ci)dt . ................. (C-3)
o
The 1/; function used to model the reaction of HF
and HCI with sandstone was
1/;=
1/;slow (C
HF
) for cf> > cf>e, ......... (C-4)
where
for A < Ap
36
2
/
3
aCi ~ ) 1/3 (Di A ) 213
r(1I3)p
s
M2 2L
for A > A p' ............ (C-5)
2hakC
HF
Y;
1/;slow (C
HF
) = A 2 , (C-6)
Ps
and cf>e is the porosity after carbonates have been
removed. This 1/; function assumes an initial dif-
fusion-controlled reaction of acid with carbonates
followed by the slower reaction-rate controlled
reaction with silicates and quartz. It should be noted
41
that the 1/; fast function used in the model for A < A
is an approximation to the derived function (Eq. if-
1). This approximation was necessary because the 1/;
function proportional to A
2
causes a numerical
instability in the evolution equation. 16
As both 1/;fast and 1/;slow are separable functions of
C and A, the evolution equation can be transformed
to Eq. C-2 and solved before the acid balance
equations. This solution was made using a modified
finite difference technique,16,17 and the pore-size
distribution 1/ (7) then was stored as input to the acid
balance equations. In this form the appropriate pore-
size distribution can be obtained, given a value of 7.
To obtain 7, the time dependence of the acid con-
centration at a point must be established and the
integral defined by Eq. C-3 evaluated.
The initial condition needed to solve the evolution
equation is the initial pore-size distribution 1/(A,x,O).
For the Berea sandstone studied, the original pore-
size distribution was obtained from mercury in-
jection data taken by Core Laboratories Inc. A
comparison of these data with those obtained by
Reigle
20
in studies on several sandstones indicated
that the pore-size distributions differed little for the
various sandstones. Thus, it is felt that the initial
pore-size distribution of Berea sandstone is an ap-
propriate initial condition when studying many other
sandstones as well.
Once the pore-size distribution has been
established, the acid balance equations (Eqs. 2 and
42
3), the A
clay
equation (Eq. 14), and the 7 integral
(Eq. 22) all must be solved simultaneously. This
solution was accomplished using a previously
developed program
21
for solving coupled partial and
ordinary differential equations. The program em-
ploys orthogonal collocation to generate a system of
ordinary differential equations, which then are
solved using Gear's method
22
for stiff systems. The
complete details and computer program are
presented elsewhere. 19
SI Metric Conversion Factors
bbl x 1.589873 E-Ol = m
3
cu ft x 2.831 685 E - 02 = m
3
of
CF - 32)/1.8
=oC
ft x 3.048* E-01 m
gal x 3.785412 E-03 m
3
in. x 2.54* E+Ol mm
Ibm X 4.535924 E-01 kg
Ibm mol x 4.535924 E-01 kmol
mL x 1 cm
3
psi x 6.894757 E-03 MPa
sq ft x 9.290304* E-02 m
2
"'Conversion factor is exact.
SPEJ
Original manuscripl received in Sociely of Pelroleum Engineers office May
19, 1977. Revised manuscripl received Dec. 27, 1979. Paper accepled for
publicalion Aug. 13, 1980. Paper (SPE 6607) firsl presenled al Ihe SPE 1977 Inll.
Symposium on Oilfield and Geolhermal Chemislry, held in La Jolla, CA, June
2728.
SOCIETY OF PETROLEUM ENGINEERS JOURNAL

You might also like