Waterflood Surveillance in

the Kuparuk River Unit With

Computerized Pattern Analysis
L.R. Chapman, SPE, and R.R. Thompson, SPE. Area Alaska Inc.

Summary. This paper describes the computer-aided waterflood surveillance method used in the Kuparuk River field. !he flood is
monitored pattern by pattern, with production and injection allocations verified by comparison of measured ~eld pressures with pressures
calculated by material balance. The resulting detailed knowledge about watert100d progress has led to operatIOnal changes and workovers
to optimize watert100d performance.

Introduction
The Kuparuk River Unit lies on the North Slope of Alaska about formance to date without attempting to predict future performance.
40 miles [64 km] west of the Prudhoe Bay Unit. The Kuparuk The method presented represents a simpler, lower-cost alternative
reservoir contains some 4.4 billion bbl [700 x 106 m 3] of original to the use of simulation as a watert100d evaluation tool.
oil in place (OOIp).l A combination of relatively low API gravity
(24 API [0.91 g/cm 3]) and an initial undersaturated condition
0
Allocating Fluids by Mat...lal Balanc.
makes watert100ding this reservoir a necessity. To complicate Fig. 1 is a schematic of a typical Kuparuk watert100d pattern made
matters, the Kuparuk is highly faulted in some areas and is divided up of two sands with differing flow characteristics. To monitor and
into a lower body called the A Sand and an upper body called the optimize watert100d performance in each sand, factors are needed
C Sand, with the C Sand considerably more permeable. I to allocate production and injection vertically between these two
Because of the large reserve stakes and the geologic complexity sands and area1ly among patterns. Separate sets of allocation factors
of the Kuparuk, a comprehensive watert100d surveillance effort is are used for oil, gas, and water.. .
necessary. A pattern-by-pattern knowledge of performance is needed Vertical allocation factors can be determined fairly accurately
for waterflood optimization through workovers, infill drilling, and and at a moderate cost with spinner surveys in injection wells. The
other operations. This knowledge is also important when selecting determination is much less accurate and very costly in producing
EOR candidate areas in the watert100d. wells, particUlarly after water breakthrough occurs. An additional
Because of the large amount of data involved in any watert100d problem is caused by the significant changes in well profile that
project, a computer can be helpful to engineers involved. A question can occur between spinner surveys. Historically, little or none of
arises about the actual amount of "help" the computer should give. this type of data may have been taken in a mature field, hampering
The program can range from a simple production data base to a efforts to evaluate pattern-by-pattern t100d performance.
comprehensive full-field reservoir model that is used to track the Various methods can be used to determine areal allocation factors.
waterflood. The more automated a computerized method is, the less Injection~well tracer tests can aid in estimating areal allocations,
work the engineers have to do. By being involved less, however, but tracer tests are costly and give mostly qualitative results. The
engineers do not become as familiar with the detailed t100d per- simplest method is to assume a homogeneous, constant-thickness
formance and the program may not be as flexible as a program re- reservoir and to base the factors geometrically on angle open to
quiring more hands-on data manipulation. Also, a complex computer t1ow. 2 Although simple to calculate, geometric allocation factors
program tends to be a "black box" that only one or two people (even after adjustment for nonideal conditions) may not be an ac-
may understand fully. The computer program described here in- curate representation because of areal variations in reservoir prop-
tentionally requires a great deal of manual manipulation by the en- erties, such as porosity and pertneability. Reservoir. discontinuities,
gineers. such as truncations and faulting, can also vary the allocation factors.
The cornerstone of a sound surveillance effort, computerized or Areal allocation factors also need to be adjusted over time to match
manual, is a good allocation of production and injection to every field performance as different patterns and sands pressure up and
zone within each waterflood pattern. Allocation factors can be deter- eventually reach breakthrough.
mined or estimated in a variety of ways from field data or reservoir The methOd described here handles t1uid allocations by careful
parameters. Most ofthese methods have large uncertainties and do study of each pattern and uses a material-balance method to aid
not track t1uid allocations as they change over time with varying in correcting the allocation factors. Typical material-balance ap-
reservoir conditions. plications involve. the use of measured pressures and production
The method outlined in this paper uses a material-balance tech- data to determine the original volume of oil in a reservoir. In this
nique to backcalculate a pressure based on the production and in- waterflood surveillance method, material balance is used in reverse.
jection allocated by pattern. These pressures, calculated monthly, The OOIP of a pattern is determined volumetrically, then production
are compared with actual field pressure data to check the pro- and injection are allocated into a pattern. Next, material-balance
duction/injection allocation factors, which are adjusted manually. equations are used to calculate what the pressure should be, given
After the t1uids have been adequately allocated to each pattern and the known reservoir volume, PVT relationships, and allocated pro-
zone, helpful output, such as pattern performance and conformance duction and injection. The calculated pressures are then compared
plots, are plotted by the computer. This leaves the interpretation with measured values to check the allocation factors.
to the engineers involved, getting them much closer to the problem. The general form of the material-balance equation is 3
The watert100d surveillance program is used to look for anomalies
in field performance, such as low volumetric efficiencies or poor N(Bt-Bti)+FvNBti[(BgclBgi)-l]
injection/voidage ratios. More often than not, these anomalies turn + [(NBtiSwio)/(1-Swio)][(Btw - BtwnlBtwil
up opportunities for waterflood improvement.
While waterfloOd performance parameters could be calculated +[FvNBtiSwigl(l-Swig)][(Bw-Btwi)/Btwil + WeB w
on the basis of a detailed history match of a simulation model, the +GiB'g+ WiBw+[NBti/(1-Swio)+FvNBti/(1-Swig)]cjAp
main thrust of a model is to explore future scenarios with the model
used as a predictive tool. Alternatively, the method described in =NpBo+(GpsBg-NpRsBg+GpcBgc)+ WpB w' (1)
this paper is intended to provide an in-depth look at waterflood per- Because Bt=Bo+(Rsi-Rs)Bg, (2)
Copyright 1989 Society of Petroleum Engineers [NBoi/(l-Swio)]=Vp , (3)
Journal of Petroleum Technology, March 1989 277
 • •
f ...
...
Pettern Element
!
• •
Fig. 1-Typlcal Kuparuk waterflood pattern.

patterns were divided into injector-eentered elements, as shown in

'00,---....,...----------- ---,
Fig. I. Each pattern was then named after its center injection well.

Initialization. Development of the Kuparuk River field has been

80 spaced over several years, resulting in migration of fluids from un-

13----------__ developed areas to the active producing wells. To avoid having to

track the primary migration from pattern to pattern, the program
'0 -----~-
-\ is "initialized" at the start of waterflood and primary production
values are calculated on the basis of input pressure and saturation
data. Each pattern has a unique initialization time based on when
.0
\ water injection began in that pattern. The initialization process yields
~
calculated primary recoveries that reflect the conditions in each
pattern at waterflood startup, assuming that startup pressure and
Legend
20 saturations can be reasonably estimated. For this surveillance effort,
Adjusted A %

~~u
pressure is estimated on the basis of measured field pressure data
• Spinner A % and gas saturations are estimated with correlations of saturation with
o -h....,.....,....,...,.........,.......,.....,.....,....,...,.........,.......,.....,.....,....,....,.........,.......,....,.....:;o0::..s;:!:p:;:.inn::;.'...::c..::..-l7. pressure developed fro~ pattern simulation work.
SON 0 J F M A M J J A SON 0 J F M A M J J A SON 0 J F
1984 1985 1986 1987 The following equations are used to calculate effective primary
TIME production of oil, gas, and water:
Fig. 2-Vertlcal water-Injection split adjusted by program N px =N- Vpx(SoxIBox) , (6)
shown with spinner data. G px =G- VpA(SoxIBox)Rsx +SgxIBgx] +Gix , (7)
and Wpx = VpS wi - Vpx(SwxIBwx) + W ix' (8)
and for Kuparuk Bti=Boi ' Fv=O, We=O, Gpc=O, Swio=Swi' and
Gps=Gp , and assuming that Brw=B w and B'g=Bg , then Areal Fluid Allocation. The areal fluid allocations are adjusted
manually on the basis of comparisons of calculated pressures to
N[B o +(Rsi-Rs)Bg-Boil field data. Field pressure data are often cheap to obtain, particu-
larly when surface falloff tests can be performed on the injectors.
+ [(NBoiSwi)/(l-Swi)][(Bw - Bwi)IBwil The initial estimates of areal allocation factors were based on ad-
+ GjBg +WiBw + VpcfAp justed angle open to flow. When the calculated pressures deviate
from field data, the areal allocation factors are changed to obtain
=NpBo + (GpB g -NpRsBg ) + WpB w' (4) a better match, resulting in more accurate areal fluid allocations.
Next, the equation is rearranged in the following form:
Vertical Fluid Allocation. The C Sand in the Kuparuk River field
NBoi=(N-Np)Bo+[Gi+NRsi dominates well performance because of its higher permeability and
-(N-Np)Rs-Gp]Bg+(Wi-Wp)Bw natural fractures. C Sand dominance was even more pronounced
in injection wells at waterflood startup, when the C Sand was more
+ VpSwi(B w - Bwi)IB wi + Vpcft:.p. . (5) depleted than the A Sand. However, as the C Sand approaches fill-
In this form, a constant term is on the left side of the equation, up in a pattern and its reservoir pressure increases, a gradual
while the right side contains pressure-dependent terms. Within the diversion of injected fluid from C Sand to A Sand is observed,
computer, the pressure is varied until the right side of Eq. 5 matches changing the actual vertical injection allocations (splits) with time.
the left. This is analogous to pressuring up on a given amount of Assuming constant vertical splits between measured spinner surveys
fluid until it fits within a high-pressure tank of known volume. results in premature calculated C Sand fill-up and slow A Sand
waterflood progress. A vertical-split-adjustment method ha~ been
Application of Method developed to adjust the injection allocations into each sand, de-
Several enhancements were made to the basic material-balance pending on their respective pressures and saturations. Fig. 2 shows
waterflood-monitoring program described above to handle specific an example of water-injection split adjusted with this method for
circumstances at the Kuparuk River field. These enhancements, one pattern in the Kuparuk River field with good agreement with
along with other basic elements of the application of this water- field spinner measurements.
flood surveillance method, are described in the following sections. The basic principle behind the adjustment routine is the following
form of Darcy's law 3 :
Pattern Elements. The Kuparuk is currently being waterflooded
on a 320-acre [130-ha] linedrive pattern. For this analysis, the q=[0.OO7082kkr h(p-PMf)]1 {BJ.'[ln(relrw) -0.75]}. . ..... (9)
278 Journal of Petroleum Technology, March 1989
Rearranging this equation to put constant (or nearly constant) terms another split after water breakthrough is established in the second
on the left side and variable terms on the right yields sand alleviates the problem.
If no production-profile data are available to indicate when water
0.007082kh/[ln(relrw) -0.75] =qBI-'/[kr(p-Pwf)]' (10)
breakthrough occurs in the second sand, it should be possible to
When spinner-survey data are available, each zone's measured deduce such a breakthrough by observing anomalies in the pressures
fluid rates and material-balance pressure, along with the other known calculated by the surveillance program. After a breakthrough in
variables, can be specified in the right side of Eq. 10. Thus, the the second sand, if the surveillance program is still attributing all
left side of Eq. 10 can be solved for without any knowledge of ab- produced water to the sand with initial water breakthrough, that
solute permeability or drainage radius. The constant terms on the sand will have too Iowa calculated pressure because of overesti-
left are referred to in this paper as a flow index, If. For every mated voidage and high pressures will be calculated for the sand
month until the next survey, the flow rate into each zone is calcu- with the later water breakthrough because of underestimated
lated by substituting the flow index back into Darcy's law: voidage. The compensating response of changing injection splits
is not quick enough to mask the anomalous pressures caused by
qo = [/fokro(p -Pwf)]/Bol-'o, (11)
misallocated water breakthrough.
qw=[Ifwkrw(p-Pwf)]IBwl-'w, (12)
Gas Cycling Calculations. One complication of waterflood sur-
and qg =qoR s + [/fokrg(jJ-Pwf)]/Bgl-'g (13)
veillance at the Kuparuk River field is the presence of injected gas
The new calculated flow rates are not used directly but are ratioed in the reservoir. Because a very limited market is available for North
to get new vertical splits that are then applied to the patterns' actual Slope gas, the produced gas at Kuparuk has been reinjected into
total monthly production and injection. Flow indices are calculated the reservoir. Early in field life, this gas was injected only into Sand
for each pattern and sand for both production and injection. C on the peripheries of the field, the objective being storage of the
Flow indices for produced water and produced oil should theo- gas until such time as it was needed. Because the Kuparuk River
retically be equal and constant unless a well becomes fractured or field has relatively little structural relief, the highly mobile gas
damaged. However, work with pattern model output showed that, readily migrated into many producing areas, causing problems with
as calculated by this adjustment routine, the flow indices are neither high produced GOR' s and limited gas-handling capacity. Many of
equal nor constant. Oil flow indices calculated from pattern simu- these gas-affected areas are now being waterflooded. More recently,
lations varied with time, decreasing during early waterflood and gas has been injected in other areas of the field in an immiscible
then leveling off when steady-state conditions were reached. The water-alternating-gas (WAG) process designed to control the mo-
water flow indices started off much higher than the oil, then ap- bility of the gas and to improve recovery. 5
proached the oil as the rate of change in produced water cut slowed To determine volumetric efficiencies of the WAG and gas-affected
down. Several factors contribute to cause these differences and patterns properly, injected gas saturations and evolved formation
changes. Among them are the methods used to estimate fluid satu- gas saturations are tracked separately in the surveillance program.
rations near the wells, which will be discussed below, and the fact Appropriate portions of produced gas are treated as cycled injected
that the surveillance program's pressures and saturations used to gas, on the basis of gas saturations, pressures, bubblepoints, and
calculate flow indices lag behind by 1 month. To compensate for a user-entered gas-breakthrough date. The need to separate pro-
these effects, separate flow indices for produced water and oil are duced gas into formation gas and cycled gas results from the as-
used and recalculated every month. sumption that essentially no injected gas will be able to go back
Saturations used to calculate relative permeabilities governing flow into solution in the oil because of compositional effects and limited
from or into a well should not be average pattern saturations, but contact area. This affects pressure and bubblepoint pressure calcu-
saturations in the immediate area of that well. For this adjustment lations. (Bubblepoint pressure is calculated from the amounts for
routine, the effective water saturation governing injection is assumed formation gas and oil in place at any given time. Free injected gas
to be the average water saturation behind the front, as determined may be present when a pattern is actually above its calculated bub-
from techniques developed by Welge. 4 A tangent to the fractional- blepoint pressure.) The separation also enables a meaningful fill-
flow curve, going through the point where Sw equals initial water up calculation for WAG patterns that is based on filling up the in-
saturation, is extrapolated to anfw value of 1.0. The water satu- itial formation gas saturation.
ration at that point is the average Sw behind the front. The water The method of calculating cycled-gas volumes requires the user
saturation near the wellbore will be somewhat higher than the to enter a cycled-gas-breakthrough date, based on engineering
average saturation behind the flood front because of the higher water judgment and GOR behavior, if gas breakthrough occurs below bub-
throughput around the well. This assumption has given good re- blepoint pressure. The surveillance program will assume that all
sults, however, when compared with field data. produced gas is formation gas before that date. Afterward, formation
The effective water saturation for production calculations is also gas is calculated as solution gas plus a portion of the free produced
taken from Welge's work: gas. The formation gas production is assumed to be proportional
to the formation gas saturation divided by the total gas saturation
Swe=Sw-VpJo' (14) in that pattern. Cycled gas is simply the difference between total
Note that when measured splits are entered, fo is known for that produced gas and formation gas produced. When pressure is above
month, but when splits are not entered, the previous month's fo the bubblepoint in a gas-affected or WAG pattern, cycled gas is
is assumed to apply. assumed to be all gas produced in excess of solution gas. Equations
An iterative solution is used for the producers' effective gas satu- are shown below.
ration. Gas saturation is incremented until the following equation
holds true, within tolerances: . Formation gas saturation:
R=R s + (krg/kro)(Bo/Bg)(l-'o/l-'g)· (15) Sgf=Sg,-(GiipBglVp), (16)
The GOR here is known when measured splits are entered and is
assumed equal to the previous month's GOR when no new splits
are entered.
Formation GOR:
This vertical-split-adjustment routine does not attempt to predict
which zone is experiencing breakthrough, but depends on the en- Rt=Rs+(R,-Rs)(SgfISg,) (17)
gineer to enter that information, based on field data and/or engi~
neeringjudgment. Water flow indices can be calculated after water Formation gas:
breakthrough, and the adjustment routine can take over. The routine
D.Gpt=MVpRf · (18)
is not able to track the water breakthrough in the second sand very
accurately, however, because it is tied to the previous month's frac- Cycled gas:
tional flow of oil. When that fractional flow is changing rapidly,
as at initial breakthrough, misallocations are likely. Specifying D.Gpcyc=D.Gp-D.Gpf' (19)
Joumal of Petroleum Technology, March 1989 279
 5 ..........,----".--"",--/'7'""----.,.---T'OO
I / ''-J/ "--..

OJi
4
,
I

I
,/
'/
/ "
"", .0 so

, r'
8
toe
0.8

0.7
o :
,/
/ \
\
40

~ 20

g 0.6
:;>
o
3

I
/
\
,, 0
"-
3Ii:
~
l<

~ ,
·0.5 -20
;;s:;
2 '
,

I \
0.4 r,
"
-40

~
\
I
0.3 1
,
I
\
,, I
\\/ ..... _'\
,
-'0

~ 0.2 I
,
I , _ - -80
, \
I " I •
0.1 '00
1986 1987
0
0 ~ ~ ~ U ~ MUM ~
• ~LV9IQ _ x % FlLWP_
VOLUMETRIC EmCIENCY, NET VpHdi
Fig. I5-Example of program output, InJectlon/voldage ratio
Fig. 3-Example of program output, Staggs2 plot. and flll-up plot.

100 5-r-----------------,I-....... 'OO

II
80

4-1···························· ..·······················................. .........

"
",.1 +.0
I
, I
10
..
....... .. .. ,
.~ 40
~-~~

I '
.. ' ........
'
o I I
................ ~3
~
,,/20

ll'
...............................
§ I-==---_-_-------·~·---·--'-;I/l·. o
I.L ~
i
Ii:
-l ............
------=... ',,,
~~
/~
1
2 -20 l<

/It I '; I -4.0

I I
I I
-60
0.1 I
I
, I
I
-80
I ; I
Lagend
,/ -- /',
I

;"
O._--,----"'-......:...-..,----'-"--r----'-·'OO
•
II .~OIL __ 1986 1987 1988
0.01 i,l
Ii' !.l9_0J\!._~Y.!'Y~
• ~LV9IQ _ x % FlLWP_
0 0.2 0.4 0.6 0.8 1
CUM WATER VpHdi
Fig. 7-Example of poor waterflood performance as shown
FIg. 4-Example of program output, WOR va. VpHdI plot. by In)ectlonlvoldage ratio and fill-Up plot.,

Volumetric efficiency:
35
Ev=[(Wj - Wp)Bw+(Gj-Gpcyc)BgJlVpHd' (20)
30 Percent fill-up:
M=(Sgfx-Sgf)/Sgfx' ; (21)

,
25

Variations on this procedure could be used when analyzing other

~ 20
WAG processes.

~toe 15
I
Interpretation
/
/ The first step in this computerized,surveillance method is for engi-
10
/
) neers to run the program iteratively, evaluating calculated pattern
pressures against measured pressure data and adjusting allocation
5 Legend factors as appropriate. The number of iterations required for satis-
D~~ factory convergence depends on the complexity of the reservoir and
D WOD£t.RtC
0 other complicating factors, such as the existence of a gas-affected
0 0.2 0.4 0.6 0.8 I
CUll VpHdI area in the Kuparuk. Often, difficulty in obtaining a satisfactory
. match indicates a problem in a pattern, which may lead to an 0p-
FIg. I5-Example of program output, recovery va. VpHdI plot. portunity Ito improve performance. Once a set of production and
injection allocation factors has been determined that results in a
pressure history match for each pattern, results are presented in
several forms that are useful in interpreting pattem-by-pattern per-
formance.
The first of these descriptive tools is a graph of cumulative
recovery vs. volumetric sweep efficiency (conformance plot) from
Staggs,2 as ~ inFig. 3. Cumulative recovery (or recovery ef-
280 Journal of Petroleum Technology, March 1989
ficiency) is in fractions ofOOIP. while volumetric sweep efficiency These patterns may be patterns close to the bubblepoint or patterns
is defined in Eq. 20. A waterflood performance envelope is de- within a proposed EOR area. By allociu:ing water on the basis of
fined by drawing an obtuse triangle bounded by recovery at the a pattern-by-pattern knowledge of the flood status. reserve .losses
start of waterflooding. the maximum oil r~overy at E v=100%. resulting from. injection supply problems can be avoided. .
and one additional point. This third point. A. the net injection re- Results of the surveillance program have also been used as the
quired to displace the existing gas saturation at start of water- s~rting point for other studies. For example. predictions of pro-
flooding. is defined as duced water rates and water-injection needs used to justify water-
supply upgrades were initialized from program results. Evaluation
A =Sgx/(l-Swi -Sorw)' (22) of potential sites for EOR also made use of the surveillance program
Because actual performance cannot fall outside the performance and led to interim operational recommendations in the chosen areas
envelope. this plot is useful as another check on fluid allocations to help ensure that minimum miscibility pressure will be achieved
for a pattern. It can also indicate when tltere are fluid-control by the time EOR is scheduled to start up.
problems. such as thief zones outside of the producing horizon. The Kuparuk waterflood surveillance program is also being used
If changes are noted in the slope of this plot. the cause should be to help select patterns for· selective infill drilling by locating poorer-
investigated. performing patterns. once these patterns have been identified. cross-
Plots of WOR vs. VpHdi (Fig. 4) or recovery vs. VpHdi (Fig. 5) sectional studies and other geological and engineering data are used
shown with ideal curves from a generalized pattern simulation are to determine whether inml drilling would improve recovery.
useful for higWightinganomalies in performance. such as early or
late water breakthrough or poor recovery. Examination of these Conclusions
plots can help identify areas for selective inml drilling. The waterflood surveillance computer program used at the Kuparuk
Other parameters-sueh as percent fill-up and injection/voidage River field is a useful tool that has enabled detailed analysis of large
ratios (Fig. 6). pressure. and production and iqjection hfstories- amounts of data. The program enables engineers to gain a more
are plotted vs. time to give clues to pattern performance. Maps of thorough understanding of waterflood progress. Problel;ll areas are
field pressures over time help to keep the flood balanced and to pointed out so that corrective action can be taken. such as performing
monitor the pressures of potential miscible-gas EOR areas. workovers and installing downhole orifices. The general theory of
Patterns that are indicated by this analysis process to be per- the surveillance program used at the Kuparuk River field is readily
forming poorly can be compared with the better-performing patterns transferrable and applicable to many other fields and/or processes.
to identify the cause of the poor performance. Used along with ge- This procedure is especially useful in mature waterfloods that
ological information and well histories. the information gained do not have a good production and injection log history. ?yfost fields
allows the engineers involved to analyze the performance of the have an established history of reservoir pressure data. so this
Kuparuk reservoir systematically. procedure would be most useful in determining pattern performance.
If numerous pressure tests have not been run. correlations of
Results reservoir pressure to wellhead injection pressure can be used.
Program results and the engineering understanding of the water- Before an EOR process is initiated in a waterfloolied field. a de-
flood have aided in many operating decisions. leading to the op- tailed knowledge of pattern performance is mandatory. All other
timization offield performance. For example. many wells have been things being equal. the EOR process should begin in an area where
worked over to provide mechanical isolation between the A and waterflood performance was the best to reduce risk from factors
C Sands. This enables profile control by performing stimulations such as poor reservoir continuity and low overall conformance. If
in the less prolific A Sand and/or using restrictive orifices across a miscible process is to be used. this procedure can be helpful in
the C Sand to ensure that the A Sand is being flooded adequately. managing waterflood pressure to ensure miscibility. The methods
One specific example is the A Sand pattern performance shown used to handle gas injection discussed in this paper can be used
in Fig. 7. The injection well in this pattern was injecting through to continue surveillance after solvent injection startup to track the
a single completion into the A and C Sands. Minimal iqjection was miscible flood's performance.
entering the A Sand. as reflected by the poor injection/voidage ratio Nomenclature
exhibited in Fig. 7 and in spinner surveys. If A Sand injection had
continued at such low rates. waterflood reserves would have been A = net injection required to displace existing gas saturation
lost because of shrinkage of the residual oil and reduced sweep ef- at start of waterflooding. fraction VpHd
ficiency caused by increased oil viscosity at lower reservoir Bg = gas FVF. RB/scf [res m 3 /std m 3]
pressures. The injection well in this pattern was recompleted in early B' g = injected gas FVF. RB/scf [res m 3 /std m 3]
1988 to a selective single completion of the type described by W 00- Bge = gas-cap gas FVF. RB/sef [res m 3 /st4 m 3]
dling. 6 C Sand injection has been hydraulically limited by small Bo = oil FVF. RB/STB [res m 3 /stock-tank m 3]
orifice valves. allowing higher A Sand bottomhole injection Bt = total oil FVF. RB/STB [res m 3 /stock-tank m 3]
pressures. A Sand injection is now replacing voidage and proceeding B tw = total water FVF. RB/STB [res m3 /stoek-tank m 3]
toward fill-up. as Fig. 7 shows. and the A Sand reservoir pressure Bw = water FVF. RB/STB [res m 3/stock-tank m 3]
is estimated to have increased by 70 psi [480 kPa] in the first 5 Cf = formation compressibility, psi -1 [kPa -1 ]
months after the workover. E r = recovery efficiency. fraction of OOIP
The need for this workover could have been identified solely from E v = volumetric sweep efficiency. fraction
injection-well spinner data. However. the surveillance program 10 = fractional flow of oil. fraction
enables us to identify and prioritize such workover candidates on F v = ratio of initial gas-cap reservoir volume· to initial
a fieldwide basis. Quantification of estimated rate of decline of reservoir oil volume. fraction
reservoir pressure is made possible by the program. aiding in es-
timating potential reserves loss. Note that program results without
G =initial gas in place, scf [std m 3]
G i = cumulative gas injection. sef [std m 3]
field data to back them up would not be used to justify a major
Giip = cumulative gas injection in place. sef [std m 3 ]
expenditure. such as a workover. because the solution provided by
the surveillance program is not necessarily unique. The value of Gp = cumulative gas production. scf [std m 3 ] ..

the program for evaluating workovers is in higWighting potential Gpe = cumulative gas-cap gas produced. scf [std m 3]
3
Gpcyc = cumulative cycled injected gas produced. sef [std m ]
areas for additional data gathering and in quantifying potential 3
benefits and ranking projects after workover results have been Gps = cumulative solution gas produced. sef [std m ]
verified by field data. i1Gp = gas produced in 1 month, sef [std m 3]
Day-to-day operational recommendations. such as allocation of i1Gpcyc = cycled injected gas produced in 1 month. sef [std m 3 ]
limited water injection among injection wells. are aided by evalu- i1Gpf = formation gas produced in 1 month. scf [std m 3]
ation of program results. When injection-water supply problems h = formation thickness, ft [m]
arise. water is allocated to patterns most in need of injection water. If = flow index

Journal of Petroleum Technology. March 1989 281

 k - permeability,' md Subscripts
kr = relative permeability, fraction g = gas
. M = fill-up, % . i = initial
N = initial oil in place, STB [stock-tank m3] 0= oil
Np = cumulative oil produced, STB [stock-tank m3] t = total
f),Np = oil produced in I month, STB [stock-tank m3] w = water
p - average reseJ;Voir pressure, psig [kPa] x = at waterflood startup
Pw/ = flowing bOttomhole pressure, psig [kPa]
I1p = change in reservoir pressure, initial to current
conditions, psi [kPa] Acknowl4tclgment
qg = gas flow rate, sefID [std m3 /d] We thank Arco Alaska Inc. for permission to publish this paper.
qo = oil flow rate, STBID [stock-tank m3 /d]
qw = water flow rate, STB/D [stock-tank m3 /d] Referenc._
re i::: radius of boundary, ft [m]
1. Carman, G.I. and Hardwick, P.: "Geology and Regional Setting of
;Ow =' wellbore radius, ft [m]
Kuparuk Oil Field, Alaska," AAPG Bull. (1982) 67, No.6, 1014-31.
R = produced OOR, sef/STB [std m3/stock-tank m3] 2. Staggs, H.M.: "An Objective Approach to Analyzing Watertlood Per-
~f = produced formation OOR (excludes cycled injected formance," Proc., Annual Southwestern Petroleum Short Course,
gas), sef/STB [stci m3/stock~tank m3] Lubbock, TX (1980) 133-50.
R s = solution OOR [std m 3/stock-tank m3] 3. Amyx, I.W.; Bass, B.M., and Whiting, R.L.: Petroleum Reservoir En-
S = saturation, % gineering, McGraw-Hill Book Co. Inc., New York City (1960) Chaps.
2 and 8.
Sgf= formation gas saturation (excludes injected gas), % 4. Craig, F.F. Ir.: The Reservoir Engineering Aspects o/Waterflooding,
Sorw = residual oil saturation t<1 water Monograph Series, SPE, Richardson, TX (1971) 3, Chap. 3.
Sw = average water saturation, fraction 5. Champion,. I.H. and Sheldon, I.B.: "An Immiscible WAG Injection
Swe = effective water saturation at producer, fraction Project in the Kuparuk River Unit," paper SPE 16719 presented at the
Swig = initial water Saturation in gas cap, % 1987 SPE Annual Technical Conference and Exhibition, Dallas, Sept.
Swio = initial water saturation in oil zone, % 27-30.
6. Woodling, G.S.: "Recompletion Workover Program at the Kuparuk
Vp = PV, RB [res m3] River Field," paper SPE 16932 presented at the 1987 SPE Annual Tech-
VpHd = displaceable HCPV, dimensionless nical Conference and Exhibition, Dallas, Sept. 27-30.
VpHdi = displaceable HCPV injected, dimensionless
Vpi = PV injected, fraction .
We = eumulatiye water iilflux, STB [stock-tank m3]
WI :i= cumulative water injection, STB [stock-tank m3)
Wp = cumulative water production, STB [stock-tank m3] . Original SPE manuscript received for review March 23, 1988. Paper accepted for publl·
cation Dec. 2, 1988. Revised manuscript received Nov. 21, 1988. Paper (SPE 17429) first
p. = viscosity, cp [mPa's] presented at the 1988 SPE California Regional Meeting held In Long Beach, March 23-25.

282 Iournal of Petroleum Technology, March 1989