You are on page 1of 10

SPE

SPE 21760

Design and Implementation of the First Arctic Offshore Waterflood,


Endicott Field, Alaska
G.R. Adamson, H.L. Hellman, and R.R. Metzger, BP Exploration (Alaska) Inc.
SPE Members

Copyright 1991, Society of Petroleum Engineers Inc.

This paper was prepared for presentation at the Western Regional Meeting held in Long Beach, California, March 20-22, 1991.

This paper was selected for presentation by an SPE Program Committee follOWing review of information contained in an abstract submitted by the author(s). Contents of the paper,
as presented, have not been reviewed by the Society of Petroleum Engineers and are subject to correction by the author(s). The matenal, as presented, does not necessanly reflect
any position of the Society of Petroleum Engineers, its officers, or members. Papers presented at SPE meetings are subject to publication review by Editorial Committees of the Society
of Petroleum Engineers. Permission to copy is restricted to an abstract of not more than 300 words. Illustrations may not be copied. The abstract should contain conspicuous acknowledgment
of where and by whom the paper is presented. Write Publications Manager, SPE, P.O. Box 833836, Richardson, TX 75083·3836. Telex, 730989 SPEDAL.

ABSTRACT INTRODUCTION

The Endicott Field is the first Arctic offshore producing oil Endicott Field is located two miles offshore in the Arctic
field. Early reservoir studies showed that optimum OCean, about 8 miles (12.8 km) east of Prudhoe Bay Field.
recovery would require start-up of waterflooding within the To date, 74 wells have been drilled from two man-made
first two years of production. This paper describes the gravel islands, both constructed in 14 feet (4.3 m) of water.
engineering and geological work done to implement and The main production island (MPI) contains production
optimize the Endicott Field waterflood. facilities and about one-half of the wells. The remaining
wells have been drilled from the satellite drilling island
The reservoir at Endicott Field is contained within the (SOl).
Mississippian Kekiktuk Formation. These multistory fluvial
sandstones are characterized by wide variations in rock The field produces from the Mississippian Kekiktuk
properties. Thick areally extensive shales combined with Formation. Current estimates of total hydrocarbon
sealing intra-reservoir faults, serve to subdivide the field volumes are 1.1 billion barrels (175X10 6 m3) of oil in place,
into seven distinct reservoir management sub-zones. 400 BCF (11 X10 9 m3 ) of gas cap gas, and 800 BCF
The design and implementation of this waterflood project (22X109 m3) of solution gas. Production start-up began in
extensively used simulation studies. These studies October of 1987, water injection began in February, 1988
showed that waterflooding would nearly double the with the field wide injection project completed by mid
recovery and there were significant benefits for returning 1990. To date, the field has produced over 100 MMBO
the field to original pressure. Simulation models optimized (15.9X10 6 m3) at an average daily offtake rate of 100,000
both the number and placement of production and barrels of oil per day (15900 m3/d). Reservoir pressure is
injection wells. maintained to optimize field recovery through gas
reinjection into the original gas gap and water injection.
The first production facility sealift included waterflood
facilities. This enabled a pilot program to be implemented ENDICOU GEOLOGY
within five months after production start-up to shake down
facilities, determine well injectivity and obtain early Regional Setting
waterflood performance data.
The Kekiktuk Formation is the earliest member of the
Mississippian aged Endicott Group and unconformably
References and illustrations at end of paper. rests on older upper Devonian deformed metasedimentary
basements rocks. Conformably overlying the Endicott

103
DESIGN AND IMPLEMENTATION OF THE FIRST SPE 21760
ARCTIC OFFSHORE WATERFLOOD,ENDICOTT FIELD, ALASKA

Group is the transgressive marine Itkyliariak Formation. further subdivided by the presence of a regionally
(Figure 1). extensive correlative lacustrine shale sequence.

The tectonic history for the Endicott Group is complex. Zone 3, the uppermost part of the reservoir is composed of
Numerous unconformities, syndepositional faulting during stacked fluvial channel sand sequences with
the Mississippian, and post-Mississippian tectonic carbonaceous mudstones and fine sands/silts deposited
overprints (wrenching and reactivation of earlier tectonic in interchannel floodplanes. Within Zone 3, several
elements) occur associated with the opening of the regionally extensive and correlative coals. shales. and
Canada Basin to present day North. shaley sequences (3B) further serve to subdivide the
reservoir into vertically isolated hydraulic units.
Deposition of the Kekiktuk sands occurred within
depocenters controlled by uplifted basement terrains
during Ellesmerian time. Gradually, southerly flowing fluvial
systems developed eroding basement highlands. Braided In descending order, the fluids at Endicott consist of a gas
and meandering streams deposited their bed loads over cap, oil leg, "intermediate" aquifer, tar mat, and regional
pre-Kekiktuk swamplands and low-lying basement terrain. aquifer (Figure 1). The oil column is about 340' thick,
bound by a planer gas-oil-contact at 9855' (3004 m)
Structure lIrap) TVDSS and an oil-intermediate aqUifer contact at 10,195'
(3107 m) TVDSS.
The Endicott Field is a combined stratigraphic/structural
trap. Hydrocarbons are accumulated in a southwesterly The presence of an areally extensive undulating tar mat
dipping fault block bound on three sides by major normal separating the intermediate and regional aquifers is of
faults (Figure 2). Structural dip closure occurs to the major importance in its' effect on waterflood planning and
southeast. On the northeast, the Tigvariak fault juxtaposes injection well placement. The tar mat, combined with the
older basement rock against Kekiktuk sands. To the presence of major field bounding faUlts, effectively isolates
southeast, the Mikkelsen Bay Fault juxtaposes younger the regional aquifer below 10,400' (3179 m) TVDSS from
Kayaklltkyliariak sediments with the reservoir. The Niakuk providing pressure support.
Fault, a major North Slope tectonic feature. juxtaposes
Triassic and Cretaceous sediments against the Kekiktuk. DEVELOPMENT OF A DETAILED GEOLOGIC MODEL

Seal is formed both by the conformably overlying marine Complexities resulting from the combined effects of
Itkyliariak and by regionally continuous, lower Cretaceous vertically segregated subreservoirs, areal differences in
shales which unconformably overly tilted Kekiktuk zonal rock properties, structure, and tar presence, led
antiformal fault blocks. Hydrocarbons are thought to have early-on to the realization that optimum field planning and
been predominantly sourced from the overlying development needed a detailed geologic model. Based
Cretaceous shales. 1 on previous experience gained at Prudhoe Bay Field,
reservoir description sensitivities such as sand continuity,
Several intrareservoir normal faults have been identified. shale distributions, and faulting were identified as having
These faults directly affect waterflood, field development, significant impact on partial and full-field simulators, well
and offtake planning and will be further discussed in a later completions, and reservoir management strategy. A
section. focused effort was undertaken to acquire, store, and
interpret data. These data included: a 3-D seismic survey
Stratigraphy covering fifty-five square miles (142 km 2) at 110 foot (33.5
m) spacing, detailed core and lithofacies analysis from 8
Within the Endicott Field, three major lithostratigraphic wells covering all reservoir sub-zones, dipmeter analysis of
units are defined within the Kekiktuk Formation. Figure 3 cored and uncored sections, full suite resistivity and
shows the average zonal rock properties. The lowermost porosity open-hole logs for all development wells, and
interval, Zone 1, is mainly comprised of shale and coal cased hole pulsed and compensated neutron logs where
swamp sediments with minor poor-reservoir quality sands. necessary to confirm reservoir continuity based on gas and
Zone 1 is "capped" by a major correlative coal. water movement.

Zone 2 sands are predominantly medium to coarse grained Pressure data in the form of build-Ups, statics and open-
and often conglomeratic, interpreted as braided stream hole repeat formation pressures were also critical in
deposits. These sands comprise some of the highest shaping a working reservoir description model.
grade reservoir rocks found on the North Slope with
enhanced secondary porosity as high as 28% and
permeabilities often as high as several darcies. Zone 2 is

104
seE 21760
SPE 21760 G. R. ADAMSON, H. L. HELLMAN, R. R. METZGER 3

GEOLOGIC CONTROLS ON WATERELOOD PLANNING achieve optimal well placement with respect to sand
development. To date, static and open-hole repeat
Sealing Fauns formation pressures have shOwn increased pressure
connectivity of these lower netgross sands than originally
The Endicott reservoir is broken by numerous expected. Pressure connectivity will lead to higher oil
intrareservoir fauns of minor throw. These fauns generally recovery through enhanced waterflood efficiency and
are subparallel to major field bounding faults. Six months potentially decrease the number of wells necessary to
after production start-up, static pressure data showed an adequately exploit 38/3C reserves.
anomalous pressure differential near the Midfield Fault
system (Figure 4). Average pressures for subzones 28 RESERVOIR SIMULATION
and 2A showed a 150-200 psia (1030-1380 Pal differential
across the faun, clear proof of limited transmissibility across Description of Models
the fault system. The lack of pressure communication is
attributed to the combined effects of "smearing" of ductile The development process of the Endicott Field has used
beds and possible mineralization along faun planes. Poor reservoir simulation extensively. Reservoir simulation has
pressure support in the SOl area is caused by limited evaluated well density, lateral well standoff from the gas
access to the MPI gas cap across the Midfield Fault and cap and aquifer, and waterflood benefits and timing.
ineffective aquifer support resulting from the presence of
tar. The simulation model (VIP) used for most of these studies
is a four phase partial compositional program. VIP allows
The discovery that the Midfield Fault was sealing two gas phase compositions, although all development
necessitated treating the reservoirs on each side of the simulation of the Endicott Field used the program simply as
fault system separately. Initially, this reqUired reducing a black oil simulator.
offtake on the SOl side due to rapidly falling reservoir Three types of simulation models have used: 1) full field
pressure. It was also necessary to accelerate the drilling of models (FFM's)j 2) partial field models of isolated reservoirs
water injectors to provide pressure support. of the field (PFM's) and 3) very finely gridded generic
models (mechanistic).
Sub-zone Approach to Deyelopment planning
The first FFM had coarse grids representing areas of 40
Distinct differences exist in depositional environment, acres (16 hectares). Recent improvement in computer
reservoir quality, and netgross sand ratios between sub- memory size has allowed development of a FFM with areal
zones 38/3C, 3A, and 28/2A. This has led to fundamental grids as small as 5 acres (2 hectares). These models were
differences in development planning for each sub-zone. used to optimize facilities design, estimate total
Effects of structural dip, aerially continuous coals/shales development well counts, and evaluate field wide
(vertical segregation), and a sealing Midfield Faun have development strategies such as waterflood benefits and
worked in concert to produce six separate main timing. Full Field Models use pseudo relative permeability
subreservoirs (Figure 5). The seventh subreservoir is the curves derived from fine grid cross section models.
38/3C in the down-dip peripherial edge of the field.
Partial Field Models have finer gridding than the Full Field
In the high net:gross sand subzones 3A and 28/2A, the Models. Areal gridding has been from 1 to 5 acres (.4 to 2
presence of down-dip tar and structural error were the hectares). Reservoir sand sequences are modeled with 10
critical risk factors in water injection well placement. Tar foot (3 m) layers at the top and bottom and enough
occluded sand porosity was observed to have little additional layers so that a maximum layer thickness of 50
injectivity, and there was little evidence of aquifer support, feet (15 m) is not exceeded. Reservoirs modeled using
indicating that communication through the tar was very low. PFM's are effectively hydraulically isolated from each other
Injection wells had to be placed to have access either to and require simple boundary conditions. The small
the intermediate aquifer above the tar or where no reservoir volume evaluated by these models allow fine
intermediate aquifer was present as near as possible to the gridding and improved detail.
down-dip limit of the light oil wedge (Figure 6).
Mechanistic models use very fine layering and gridding to
In the lower netgross 3C sand sequence, development evaluate optimum perforation placement, pulse tests,
planning took a different approach. Although the risk of tar single well tracer tests, and other generic concerns.
occurrence and structural error was still present, an added Mechanistic models impacted basic development
risk was the absence of reservoir quality sands. An decisions, but have had little affect on waterflood
intensive reservoir description effort was undertaken which development planning.
redefined subzone 3C stratigraphy into five depositional
sequences (Figure 3) with mappable channel trends. The Full Field and Partial Field Models have been initialized
Development wells and water injectors were drilled to with reservoir properties obtained from a detailed static

105
4 SPE 2116 0 DESIGN AND IMPLEMENTATION OF THE FIRST SPE 21760
ARCTIC OFFSHORE WATERFLOOD,ENDICOTT FIELD, ALASKA

geologic model. This static model contains up-to-date reduced recovery in the Zone 3A reservoir. These studies
reservoir geology interpretation in a digitized format. Data led to the decision not to inject gas into the upper 3B and
can be extracted in any predetermined areal and vertical 3C reservoirs to leave more gas available for injection into
grid size for use in a simulation model greatly enhancing the Zone 2 reservoirs.
the ease of building reservoir simulation models.
Model initialization and history matching is a joint geological
Modelljng Results and engineering effort. These interactions are mutually
beneficial. The geologists identify potential
Early development strategy for the Endicott Field utilized heterogeneities that could affect how to history match a
various reservoir simulation studies. Reservoir properties model. Conversely, the results of the modeling effort can
for these studies came from limited data obtained from give indications of reservoir characteristics not previously
exploratory and delineation well logs and cores. The most identified.
significant of these studies used a Full Field Model with a
38 x 10 x 24 grid with each areal grid representing 40 acres WATERELOOD IMPLEMENWION
(16 hectares). This study used Sohio's (now BP) 3D3P
simulation program. Although the areal gridding was Since field wide water injection was planned to start within
coarse, the layering was fine enough to allow thin (10ft (3 the first two years of field production, water injection
m)) layers at the top of each sub-zone to account for gas facilities were included in the production modules sealifted
underrunning shales. This model evaluated the benefits to Endicott from New Iberia Louisiana arriving in August,
and best timing for waterflooding and calculated an 1987. Priority was placed on commissioning of the
increase over primary recovery between 50 and 100%. production facilities which were completed with production
This same study estimated that waterflooding could be starting on October 3, 1987. Commissioning of the water
delayed by as much as two years after field start-up without injection facilities followed with completion in February
reducing recovery. However, it would be necessary to 1988.
repressure the reservoir to avoid any losses. Facility sizing,
well spacing, offtake rate, tubing size, and production Injection Testing
scheduling were also evaluated with this model.
Injection tests were conducted on three wells during the
After field development began, waterflood strategy was summer and fall of 1987. Injection rates into the wells were
further optimized using Partial Field Models. A separate much less than predicted and a concerted effort was made
model was created for each reservoir sub-zone on each to identify the cause. Two major factors were identified.
side of the Mid-Field Fault. These models evaluated the These were inadequate wellbore cleanup due to lack of
optimum number and location of injection wells, the best production and poor oxygen scavenging of the seawater
injection strategy and the benefits of any additional resulting in corrosion products causing plugging. Tests
production wells. These partial field models showed were suspended until the production facilities were
significantly greater recovery benefits (3-6% OOIP) for available for high rate well cleanup and until the water
repressuring the reservoir than calculated in the first Full injection facilities were available to provide oxygen
Field Model (-1% OOIP). scavenged water.
Another major effect on field development from the PFM Injection facilities
studies was a reduction in the number of water injection
wells. Simulation calculations showed that two injection Sea water and produced water injection facilities are
wells on each end of the separate reservoir sub-zones provided within the Endicott production facilities as seen
yielded better recovery than three injectors spaced evenly in figure 7. Seawater is supplied through a seawater
across these reservoirs. Recovery is increased with intake system from a water depth of 14 feet (4.3 m). The
maximum standoff between the injectors and producers. intake system includes marine life screen and bypass
These simulation studies saved the cost of drilling three to systems to exclude marine life from the intake. The intake
six wells, which cost about $3.5 million each. water is filtered through a set of downflow graded bed
filters, deareated in a gas stripping deareater and pumped
Simulation studies have influenced other operating
strategies. On the MPI side of the field, the reservoir will be to injection pressure of about 2500 psi (17X10 6 Pal via a
depleted using a combination of waterflood and gas gas turbine driven centrifugal pump. Only a single train is
displacement drives. Simulation studies were very critical provided with a nominal capacity of 140 MBPD (22300
in determining the optimum balance between water and m3/d).
gas injection. Gas produced from all wells in the field must
be injected into the gas cap of MPI sub-zones 3A, 2B and Produced water from the production train is sent through a
2A. Studies showed that over injection of gas (Le., pair of corrugated plate interceptors to remove the oil to
injecting more gas than produced from a particular less than 25 ppm. The produced water is then fed to a
reservoir) improved recovery in the Zone 2 reservoirs but pump identical to the sea water pump. The injection

106
SPE 21760 G. R. ADAMSON, H. L. HELLMAN, R. R. METZGER
SPE 21760 5

facilities are configured to allow either injection pump to overinject during the winter months to accommodate net
handle sea or produced water. voidage.

waterflood Pilot The need to over inject during the winter caused several
changes to be made to the facility design and operating
Because of the early injectivity problems encountered, it strategy. The most important change involved a pilot
was decided to implement a pilot project as soon as injection test of commingled sea and produced waters.
practical in order to determine injectivity parameters, shake Commingling the two waters in the winter months allows a
down the equipment and obtain early reservoir large increase in water injection rates by using the
performance data to aid in planning the full waterflood. A produced water injection pump to inject the commingled
single producing well was converted to water injection water. There was a major concern about the formation of
service as the facilities were commissioned. This well had BaS04 scale in the commingled water. The sea water has
been used in the earlier injectivity test program and had free ion concentrations of Barium while the produced water
been producing for about 4 months since field start-up. contains high concentrations of sulfate ions. To date,
With adequate pre-production and de-oxygenated results from the commingling tests have shown little
seawater, this well performed on injection as expected. An problem with scale formation.
additional 2 wells were added to the pilot and they also
performed satisfactorily. Pressure fall off , injection step The reduced availability of injection water during the
rate and injection profiles were performed on the wells to summer led to an unusual use for one well. Simulation of
determine fracture gradient, monitor fracture growth and the 3A subzone estimated that a producer near the
determine injection profile. Radioactive tracers were also location of a 2B injector would recover an additional 0.6
injected into the wells to monitor water breakthrough. MMSTBo (95000 stock-tank m3 ). However, these
calculated reserves would not justify the expense of drilling
The information gathered was used in planning the an additional well. Simulation of the 2B subzone showed
implementation of the full field waterflood. Early operation that relatively small volumes of injection water were
of the facilities also provided for improvement in operations required for optimal 2B recovery. From these studies, the
and pointed out the difficulties to be encountered during decision was made to alternately use the well as a 3A
the summer open water season. producer during the summer and a 2B injector during the
winter. This well has a packer tailpipe and a sliding sleeve
Fullfjeld Waterflood installed to allow isolation of each reservoir. The well has
sufficient injectivity to meet annual injection needs during
Pressure declines experienced in the SOl area caused the winter months.
implementation of full waterflood into Zone 2 and Subzone
3A sands in the main producing area to be accelerated and FIELD PERFORMANCE TO DATE
completed by the end of 1988. The addition of peripheral
areas and upper Zone 3 followed and is now complete. Although pilot injection began in early 1988, full
Well performance including injectivity and profile control development of the waterflood was not finished until mid
has been acceptable with only minor remedial work 1990. There has been minimal well response to date to
required. fully evaluate waterflood performance. However, well
response observed to date is consistent with model
Large potential benefits of repressurization have greatly predictions.
affected the operation of this waterflood project. Although
the sea water injection facilities were nominally designed Radioactive tracer has been injected in 12 water and 3 gas
for 140 MBPO (22300 m3/d) and have injected over 160 injection wells. A systematic well sampling program is in
MBPO (25 500 m3/d) on occasion, annual average rates place to insure timely observation of breakthrough.
have not yet exceeded 100 MBPO (15900 m3 /d). The Produced water from these wells is also analyzed for ion
Endicott Field is located near the mouth of the content to differentiate sea water from reservoir water.
Sagavanirktok River. High discharge rates during spring Injection water breakthrough has been observed in 10
run off cause very high levels of suspended solids in the wells, confirmed by either radioactive tracer recovery
ocean near the sea water intake for the injection plant . and/or ion analysis. Most of the observed breakthrough is
High solids levels cause frequent injection restrictions or consistent with model predictions, with the few cases of
shutdowns of the plant because of filter plugging. This early breakthrough explained by apparent high
problem can also be especially acute in the fall when conductivity through mapped faults. This type of
seasonal storms churn the ocean. Many modifications surveillance work is critical, as aqUifer water has broken
have been made to the facilities which have resulted in through in several wells unrelated to waterflood. Most non-
some improvement. The addition of a clarifier was studied waterflood breakthrough is the result of oement channels
but the investment could not be justified. To compensate or aquifer movement along fault planes. Only about 25%
for the low summer injection, it has become necessary to of the 40 MBWPD (6400 m3 /d) currently produced in the

107
DESIGN AND IMPLEMENTATION OF THE FIRST SPE 21760
ARCTIC OFFSHORE WATERFLOOD,ENDICOn FIELD, ALASKA

Endicott Field comes from waterflood breakthrough.

A few wells have shown major decreased gas production in


response to water injection. Figure 8 shows the GOR
history of well K-16 and the dramatic decrease in well GOR
in response to water injection. GOR response in most
wells is more difficult to evaluate as these wells vary in
offtake and are rate sensitive to gas production.

Continued attentive monitoring of field performance is the


key to increased recovery from the Endicott Field.
Reservoir simulation will playa major role in this surveillance
effort. A continued joint effort between reservoir
modeling, production operations and geology groups is
necessary to obtain maximum benefit for Endicott Field
development.

CONCLUSIONS

1. Early accurate and detailed reservoir description was


essential in implementing the Endicott waterflood.

2. Early reservoir surveillance pressure information is


directly responsible for modifying reservoir description
and waterflood planning.

3. A mUlti-disciplinary team approach is necessary to


ensure successful waterflood planning and
implementation.

4. Reservoir simulation will enhance the design process


and lower costs in developing and producing new oil
fields.

NOMENCLATURE

TVDSS = True vertical depth sub-sea


OOIP= Original oil in place
MBWD= Thousand barrels per day

ACKNOWLEpGEMENT

The authors wish to thank the management of BP


Exploration and the working interest owners of Endicott
Field for permission to publish this work. We also
acknowledge present and previous co-workers for their
significant contributions to the Endicott Project. We also
thank Lynda Stamper for preparation of the manuscript.
The work presented in this paper reflects the
interpretations of BP Exploration and not necesarily those
of the other working interest owners.

REFERENCE

1. Wicks, J. L., Buckingham, M. L., and Dupree, J. H., in


press, "Endicott Field - Northern Alaska Offshore
Beaufort Sea", American Association of Petroleum
Geologists Atlas of Oil and Gas Fields.

108
SPE 21760

2000 20000
1-
'I Legend
1800 1-
/ 'n. -0- a:R 1-
1600
..(
f
l'
\ • INJ - BWPD I -

1400
I \ !l
0
m
I- n r D..

~ I ,\ :::m
0
CJ)
1200
p ......, II
I
1\
-.\ n
/ \ I
/
I z
0
a: 1000 10000
0 / \ \ \ ] \/ tw
...... CJ
CD 800
...-u V~ L
....
./ -,
~
u ....
~
N
.q-

600
\ ~
\
400
\

I
I~
\/ \
• \1'\
V \ II
200
a • '"
OIL
- \ ~ I
o
/ ....... o
1989 APR JUL OCT 1990 APR JUL OCT 1991
DATE

Figure 8 - GOR history of Well K·16 vs. injection in Well L·14.

You might also like