Beruflich Dokumente
Kultur Dokumente
This paper was prepared for preeenlatlon at the Europ%an Formalwn Damage Conference held in The Hague, The Nethertends, 15-16 May 1995.
This paper was selected forpreeentstian by an SPE Program Committee following reviaw’ofinformation contained inan abstract submitted by the author(a). Contents of the paper, as presented,
have not baan reviewed by the Society of Petroleum Engineers and are subpcted to corractM by the author(s). The material, as preeentad, does not nsceeserify reflect any position of the
Sociity of Petroleum Enginaers, * offiirs, or members. Papers presented at SPE meetings are subject to pubfiiation review by Editorial Commitfeea of the Bc-Sity of Petrolaum Engineers.
,o.-L.A.
-,, !,-.”,, .. .
.“ -Vj . .“
d ,-”.
~.,++~
!-.”” t.
.. .n“. . .c,hc+~,.r
. . ... . . ..f . ~Qt ~re ~han w ~r&. lllu~t<a~o”~ Mv ~t LW COO&I. TIW ~bstram Shodd Wntam conspicuous acknowledgment of where end by whom
the pap ia pmsentad. Write Librarian, SPE, P.O. Box 633s36, Richardson, TX 75063-3B&. U.S.A. (Fecsimife 214-962-9435).
489
.
transition time. The model can be used effectively if core flow perforation into several layers, it is possible to develop an
test data is available. equationfor the flowrate vs. pressure drop.
The objective of this work is to provide a reliable and
user friendly simulator for predicting injectivitydecline in water (8)
injection wells. The simulator considers both external and where R is the total flow resistance which can be computed as
internal faltercake formation for various well completions,such the summation of the flow resistance in each layer as shown in
as openhole, cased/perforat~ ad hy~ulic~ly ~ct~ we~s” Fig. 1.The resistancein one layer @i) can be written as
The model equations used in the simulator are formulated.
Methodsfor calculatingflltmtion parametersam developedwhen Ri = (ln~-ln~i+l+l) (9)
k(ri~t)Hi &i-1 ~i+l -1
core flow test data is not available. Examples are given to
illustmte how the simulator can be run to predict the injectivity -Wke gi ~ ~xp(.wl) [10)
decline. These results can be used in conjunction with a cost ~h wi = ri/Li (11)
analysiswhich is presentedin AppendixA. Hi= ri/sinh wi (12)
ri = rp + (i-l)Ar (13a)
Li = Lp + (i-l)& (13b)
All of the model equations used in the simulator are thus the total resistanceis
listed in Table 1. There are eight sets of equationsthat have been 1 ~ r.
derived for the various types of well completions and filtration R= $.Ri+_ (14)
mechanisms.Equations for linear geometry are usually used for i=] k(rp + nAr) Lp + nAr
parameter estimation using data from core flow experiments. wherethe numberof layersaccountedfor in the summation(n) is
Modelequationsfor internaland externalfaltercake formationare determinedsuch that the radius of the n-th layer does not exceed
given separately. They can be used separately to compute the half the distance between two adjacent perforations. The second
injectivity decline for each impairment mechanism. They can term on the RHS of the above equation is the flow resistance in
also be used together to calculate the overall decrease in the radial flow region which can be computed from the radial
injectivityby using the transition time concept (l). flowequations.
In all equations, a is well injectivity and t is injection For the same injectionpressure in both the damagedand
time.A derivationof the model equationsfor linear geometryand undamagedcases, the injectivity decline (a), definedas %1%, is
openhole wells can be found in Ref. (l). Derivations of model givenby
equations for perforated and hydraulically fractured wells are
providedbelow. 1 ~q[~P+ 1]
.=&lG@
Models for Perforated Wells (15)
A perforation can be considered to be one half of an whtxeR is defti byR~s.(14)and (9).
ellipse that is rotated about its primary axis, Fig. 1. shows this External Filtration. In the case of external
e~ipsoi~ geometrywhererp is the radius of entryholeand LP is fdtration, we can assume that the external filter cake just is like
r...l:~m...:~m
C.. ~mkhm. f&\ An flri.xr rnto in 9
\UJ, u- ----- w..- -c -=.:... A .-~ -m.=mk;l~tu k- l%llnwino Ffi (14).
h ~nehtioii (k@h i%)llUW1ll& OdbllLGL a ‘kyer ul pulurwy ~c -,U p....-u.-.J ~. - “----- ---e —-l., - .,.
perforation canbe expnxsed as, the flow resistances with and without external cake formation
27rkoHpAp g +1 ale,
@= ~ /M~l
gp-1 (1) 1 ln[~ + l].&llQJ +--hn[~l ~ +1
%=
W* ~p = exp(wp) (2) iW% k-l ~p-] 1 ~ ~P-l (16)
tanh wp = rp/Lp (3) g +1
R = ~n[~]
Hp = r@nh Wp (4) * &p-l (17)
We use 13q.(1) to derive the injectivity equations for Becausethe cake thicknessis very small, we can assume that the
both internaland externalfdtration in perforatedwells. cake is concentricwith the ellipsoidal perforation (H = I-Q. The
Internal Filtration. If we view the fluid flow in a injectivitydeclinecan be definedas R& thus
perforation as radial flow around the perforation axis, then the
permeabilitydistribution at a distance of r horn the axis can be ~ ln(~
derivedfrom the fdtrationequationsin radialgeometry(1), a= P-
k(r,t) = ~
1 + ~ n(r,t) (5)
(18)
n(r, t) = kC t exp(-kn(r2-$)Wqp) (6) h an approximationwhen ~ >>rp and ~ >>rc,
& ~p=r*p+l (19a)
‘P= NPh m ~=r~p+l (19b)
Having-assumed an ellipsoidal sham for the damaged If r~= rp - k, rearrangingEq. (18) yields,
zone and discretizing the permeability distribution around the
490
SPE 30127 Shutong Pang and Mukul M. Shanna 3
1 (29)
:=l+*(ln(l+*)/ln(l +y.l) (20) _=l+Mt
c c
k,(t)
Next, we need to derive a relationshipbetweeninjection M=6f (30)
time and the thickness of the external filter cake (hc). The 6= pvclxf (31)
cumulativeinjected solid volume (Vs) at injectiontime t is f= 1- exp(-kxf/V) (32)
External Damage on Rock Face. If the fracture
V%=C\qdt (21) ennAls~tivitv/9wti11re
““.1 -””.. . .. J,-y”- ---- ic
-“ relntivelv
------- --, !~rg~ ~n~ ~h~ fQrmatiOIl
reduction,so we can use equations for linear geome~ to predict geometry is commonly available, while other parameters must
injectivitydecline. be determinedexplicitlybeforecomputingthe injectivitydecline.
Following Appendix A in Ref. (l), if the flow area is
given by A = 2 h Lf and Lf = &, V = q/A, then the injectivity Type of Well Impairment and Transition Time
declinewill be The type of well impairment internal filtration vs
a(t) = r~~ + (re-xf)] externalfiltrationhas traditionallybeen done using the 1/3 to 1/7
(27) rule. This rule of thumb states that an external filter cake is
Xf=vt, formedif the averageparticle/ Pre size ratio is greater than 1/3
for Vt < re (28a) and an internal fflter cake is formed if the ratio is less than 1/7.
Xf= re * for Vt > re (28b) In general, however, it is likely that an internal filter cake will
491
A Evaluating the Performance of Open-Hole, Perforated and Fractured Water Injection Wells SPE 30127
decline is dependent on the cake properties (permeability and and will contribute to permeability impairment. Particles much
porosity). Experimental data on cake permeability and porosity smallerthan the pore throat size are likely to be retained in pore
for some types of particles are available in the literature. bodiesand therefore,contributeless to permeabilityimpairment.
Permeabilityand porosity data for latex and silica particleswere A typical shape of the function t is shown in Figure 4.
reportedby Eylartder(3) and me listed in Table 2. Thesedata can If we assume that the permeability reduction is caused
be used directlyif the exWrimentalconditionsare similar. by pore throat biocking, the correlation between permeability
Extemai fiiter cake propertiescan ah ‘w&aimd hxn -A... :Amanu
lcuudluu
.-A RUUIIWU
*&.& nmr&.ls
yu -..”
rvmrpn~tinn
.“..”...
P21) h? w~~en &$
- -“.. .- .,- ..
direct measurementor empirical correlations.The cake porosity 1 _ z n(x,t) + (1- z n(x, t))
(42)
can be obtained if the particle volume in the filter cake is k(x,t) kp kc)
gravimetricallymeasured.The cake permeabilitycan be obtained
from flow tests if the cake thickness is measured. Empirical where kp is the permeability of the plugged pore throats and is
co~!ations such ~ the Bla.ke-Cozenyequation can be used to relatedto the ptmneabilityof the retainedparticlesthat are packed
obtain a permeability-porosity correlation for packed beds of in the pore throats.
tmticles, RearrangingEq. (42), we obtain
dz 43 b
—=l+(k- 1)Tn(x,t) (43)
kp = &.- (40) k(x,t) kP
1501-$
-A. -o ;. n In - n m An*n&fig We can express the above equation as the following
UI lhu ~au=t%FaIuu1Q=Is U. v -–V.JW WP..
WI. - --.-:... J. - . ...+ ~~ ,.I
I IE PUIUMLY
equationwhere~ is caiieddamagefactor.
the type of particle, size distribution and applied pressure.
Example data derived from Eq. (40) are shown in Table 2. The &Jl=~
resultsam very close to the experimentaldata. b 1 + ~ n(x, t)
(44)
~= ‘c(ko/kP - 1) (45)
Internal Filtration Parameters
Two parameters, the filtration coefficient (k) and the If we apply the Blake-Cozenyequation (40) to both the rock and
damagefactor (fI),are introducedin the internalfiltrationmodel. the packedparticleswith the same porosity, we obtain,
They can be determined from either experimental data or $=~d~~ (46)
empirical correlations. When core flow test data is available, k where dp is the average particle size and dg is the average grain
and ~ can be obtained from history matching the pressure drop size of the rock. Inserting Eq. (46) into Eq. (45) gives us a
data. condation for $,
The filtration coefficient is a local and time dependent
parameter.Any factors affecting the particle deposition process ~= ~ (d~/d;-1) (47)
can affect it. It is believed that the following factors are where ~ is determined from Fig. 4. Eq. (47) can be used to
imptanti (1) Pore structure (posit y, pore thrw@ody size and estimate~ if no core flow data is available.
distribution).(2) Particle size and distribution.(3) Fluid velocity. If the injected particles are much smaller than the
(4) Interactions(variousforces)amongthe pmticle,fluidand pm formation grain size, ~ + Oand most particles will be retained
surface.Variousmodels were proposedby earlierinvestigatorsto in pore bodies. The permeability impairment can then be
simulatethe process and to obtain the filtrationcoefficient. computedusingEquation(40),
Payatakes at al. (7) obtained the values of filtration ~= $S(x,t) 1-+2
coefficient from trajectory analysis based on various flow (48)
models. These values along with some experimental data are k. l-#(x,t)T
shownin Table 4.3. Both Payatakesat al. (8) and Rajagopalanet where $(x,t)=@ - (1–~) n(x,t) (49)
al. (9) also reported some experimental data for filtration
coefficients which are shown in Table 4.4. These data are
replotted in Fig. 3. The following empirical conflation between
filtrationcoefficientand particle/grainsize mtio can be obtained, A simulatorhas been written to compute the injectivity
k= f(v)* 14.481[dP/dg]0.587 (41) decline in water injection wells. The model equations listed in
Table 1 are used in the simulator.
where f(v) is the coefficient accounting for the effect of fluid The simulator is available on PCs and Unix
velocity on the filtration coefficient. The form of f(v) can be workstations.It is windowbased and menu-drivenprogram.Help
492
SPE 30127 Shutong Pang and Mukul M. Shanna 5
flow data to obtain these parameters. If no optional data is injectivitywas noted for perforated wells. This suggests that the
provided, reasonable estimates of these parameters can be effect of high fluid flow rates near the perforations needs to be
#.n-lm :*m ( ff.rh + 1 in
obtainedfromempiricaicorrelations, accoimkd for in tiie UU4UII GUG1llULG1lL
C. I-..,...
~ullvmkll
..,.,44%.:
~ .i., -. ...
--+
The output of the simulator will be the injectivity equation41). By setting f(v)= 0.1 for perforated completions a
decline,the skin factor, the depth and extent of solids invasionin Iudf-lifeof 68 days was obtained.
the near wellboreregion and the pressuredrop due to the external
filter cake. The simulator results provide not only the decline in Example 3
injectivity with time (hdf-iife of the injector) but also more This examplewas selectedto illustrate cases whereboth
detailed information about the location of the damage. This internaland externalfaltercake formationare important.The data
information is usefui in making stimulation decisions and is fromEyiander{Y):
A.a.%wh%m
uwl~llnl~
A ..1.3*JM
.dnula.wl.
tradmpntc
UWCA*... WO..O. =.. iniected
-AveraQe J particlesize dp = 12 pm,
The simulator can be used in a histo~ matching mode ParticleconcentrationC = 0.18 ppm,
to match injectivity decline for existing injection weIls.it can be Injectionrate q = 6 cc/rein.,
used to evaluate the impact of water treatment facilities on the Core permeability~ = 992 md,
economicsof a water injectionproject. Core porosity $ = 0.20,
Core length L = 5.08 cm.,
Corediameterd= 2.54 cm.
Basedon the grain size of the formationand the injected
In this section, we provide examples to show how the particle size, the transition time is calculated to be 33 minutes
simulator can be used to predict well injectivity decline for for the core experiments.The experimental data was matched to
vtious types of completions. obtain & =4.5 md, $= = 0.18, f3= 200, I = 0.035. These values
can be used to calculate the injectivity decline in open-hole and
493
.
6 Evaluating the Performance of Open-Hole, Perforated and Fractured Water Injection Wells SPE 30127
494
SPE 30127 Shutong Pang and Mukul M. Sharma 7
10, Gruesbeck, C. and Collins, R. E.: “Entrainment and As shown in Fig. Al, if tf is the time required for the
Deposition of Fine Particles in Porous Media”, SPEJ (Dec. injectivity to math as for clean water injection, then the total
1982)847-856. clean waterinjectioncost (Cf) will k
11. Todd. A. C. . et al., “Review of Permeability Damage Cf=Ct+QCP+Cs (A-2)
Studies and Related North Sea Water Injection”, (1979), where Cp is the water pumping cost Pr unit volume of water?
SPE 7883, SPE International Symposium on Oilfield and CSis the stimulation cost for recovering well injectivity.
GeothermalChemistry, Dallas, ~, Jaft. 22-% 1979. For unfilteredwater injection, there is no cost for water
treatment. If td is the life time for injectivity being as then the
total injection cost (Cd) for injection time tf will be
Cd=Q~+tf/tdC~ (A-3)
In the design and selectionof water treatmentfacilities,
Therefore,the cost ratio for clean and unfiltered water injection
the two important parameters for water quality are the particle
cartbe ObtfdM!dh ~S. (A-1) to (A-3),
size and solids concentration. Clean injection water (low solids
content and small particle size) tends to stabilize well injectivity G=l+QcdC’@ -(tdu-l)c,
but requires additional water treatment costs. In the following cd Q CP+ tdtd C, (AA)
discussion, we present a method to systematicallycomparewell If we have two different water quality requirements,the
treatment costs with water treatment costs anti obttn optiniai cost ratio for these two types of water injection wiii ‘m
wntt=r
.. —. -Wirntinnc
.=-.-..-”..”.
cf_l+Q(cu- cJ-otif-l)cs
The water treatment cost depends on the water quality
requirements. If Cu is the water treatment cost per unit volume 6- QC. +QCP+ C, (A-5)
of wa~r, c is the solids concentrationand, dp is the particle size Eq. (A4) is a special case of Equation (A-5) where Cu
in the treated water, then after injection time t, the water is equal to O(no water treatment). The net cost increase due to
treatmentcost Ct will be the injection of cleaner water (ACf = Cf - Cf) can be derived
Ct = Q CU(C, dg) (A-1) tiom Eq. (A-5),
AC. —
--, s= O
y (C..,
,.-” . Q - (~~~f~ . ~) ~a ()@
whereQ is the injected voiume and Cu is a functionof c and dg.
where the first term in the RHS is the water treatment cost
increasedue to the better water quality requirements, the second
4 Clean water iniection term is the decrease in stimulation cost due to the longer
Unfilteredwaterinjection injectionlife time.
1.0 Eqs. (A-5) and (A-6) show us how the total cost of
> waterinjectionis related to the water treatmentcost for specified
.g
“a water quality requirements, the pumping cost, the stimulation
.-8C cost and the life time of the injectionproject.
In many instancesthe decrease in injectivity is offsetby
increasingthe pumping pressure. Eventually pumping pressures
td tf willexceedtiture gradientsand the injectorwill be fractured.In
such cases the injectivity will increase without any stimulation
InjectionTime cost associated with it. This may be desimble in many cases
(low cost) and very undesirable in other cases for reservoir
Fig. Al. Schematicof water injectionscheme engineering reasons (neighboring water zones, well placement
etc. ) or for environmentalreasons.
Table 1, ModelEquations
1
–=l+Mt ~=l+Mt
a a
LinearGeometry M=6f
6= pvt/L M=~~~
kc i-$C L
f= 1- exp(-1.IJV)
495
.
SPE 30127
8 Evaluating the Performance of Open-Hole, Perforated and Fractured Water Injection Wells
Table 1. ModelEquations(continued)
1
_=l+Mt
k,(t)
M=8f
6= ~cq
2~~h$m 1$
Open hole well
f= exp(a)Ei(a) - Ei(arf/r#)l
n-1
~ .~(l~ -lnLi+l + 1) ~ln~n + 1 ln(l + a)
1 i=okr(riJ)Hi Li -1 Li+l-1 Hn ~rt-1 ~ rp - hc
—= a-l+~(
—- - 1)
a
-&& ln(l + ~
.
r -r
k(r,t) = 1
Perforatedwell 1 + ~ n(r,t) hC= ~~qdt
P
&= exp(wi)
tih wi = ri/Li
Hi= ri / sinh wi
l+CRi \
ln~)+ ~ ln(
i f i=l ri.l/ri + CRi’
1.
1 L+
—= (l-x:) a ~(!!) + ln( 1 + CR )
a ~xf) rw/Lf + CR
*–-
CR=~
Hydraulically Xf=vt for Vt C re xL~
CRi - wkfi
xf = re for Vt > re ~iko
!Yactunxl
well
~=~
1 1 + ~ n(x, t)
_=l+Mt
k,(t)
M=8f n(x, t) = O for x >Vt
496
.,
Run ExperimentalData 3) 1
%v
No. (w) (Cln/S) dp = 2.75 4.5 9.0
1
Table 5. Data Set for Examples
0.8
Injectionrate (m3/day) 1360
Formation height (m) 0.6
Wellbore/Drainagemdius (m) %00
Shot density (shots/m) 0.4
Entryradius (m) ::3
Perforationlength (m) 0.2 0.2
FmctmePermeability(d) 12.6
Fracturewidth (m) 0.005 0
Fracturehalf-length(m) 66 0 0.1 0.2 0.3 0.4 0.5
Particle/PoreSize Ratio
Fig. 4. Coefficient t vs. size ratio
497
..,.
10 Evaluating the Performance of Open-Hole, Perforated and Fractured Water Injection Wells SPE 30127
I Reauired
. Data ~ OptionalData I
I
I I
I output
(Dataand Plots) I
T*-.,* A“*.. km avmnnlt= 9
Lllpub Uala. Lsuml Waum..p” -
.2
---
0.6
0 5 10 15 20
Fig. 5. Schematicof proceduresfor Injectiontime (days)
predictinginjectivitydecline Fig. 9. Injectivitydecline due to frachue damage
— df=400, fc==.m125
~ :~
—
-—”–
open-hole
peafomted
.g “
~:s ----
___
0 100 200 300 400 500 0 100 200 300 400 500
Injectiontime (days)
InjectionTime (days)
Fig. 10. Effect of filtration parameterson
Fig. 6. Injectivity decline (Example 1) injectivity due to fracturedamage
498