SPE-63152 Waterflood Management ... Using Streamline Simulation and Traditional Waterflood Analysis
SPE-63152 Waterflood Management ... Using Streamline Simulation and Traditional Waterflood Analysis
SPE-63152 Waterflood Management ... Using Streamline Simulation and Traditional Waterflood Analysis
Waterflood Management: A Case Study of the Northwest Fault Block Area of Prudhoe
Bay, Alaska, Using Streamline Simulation and Traditional Waterflood Analysis
G. H. Grinestaff, BP-Amoco Daniel J. Caffrey, BP-Amoco
Figure 2: Prudhoe Bay Waterflood Areas like the Eastern Peripheral Wedge Zone (EPWZ) confirmed its
under-performance.
NWFB
WPWZ
FS-2
EPWZ
Field History
The Northwest Fault Block (NWFB) Area is the northwest Integrated Analysis to Address the Problem
portion of the main field area of the Prudhoe Bay Unit, see Decline Analysis
Figures 2 & 3. This structurally complex area of the field is The analysis of performance began with an examination of
bound by faults on three sides with the Eastern edge (the field performance using decline analysis. Decline curves
Gravity Drainage Waterflood Interaction Area or “GDWFI”) showed a very steep decline following the hydraulic fracture
sharing the large gas cap where gravity drainage exists. The program which occurred in 1991 and 1992 (Figure 4).
NWFB originally contained 1.75 Billion Barrels of oil.
Permeability varies significantly through the vertical section Figure 4: NWFB Oil Rate and Decline
from 10 to 2000 md.
The NWFB was produced under primary recovery (solution 1000
NWFB Area Oil Production
gas drive, gravity drainage with gas cap expansion, and a Production Decline Prior
relatively small water drive) from 1977 until August 1984 to Injection Management
when 18 patterns were converted into a nominal 320 acre
Oil Rates mbbl/day
Apr-79
Apr-81
Apr-83
Apr-85
Apr-87
Apr-89
Apr-91
Apr-93
Apr-95
Apr-97
Apr-99
Apr-01
Apr-03
80 acre well spacing. The NWFB now contains 36 injection
wells and is now a mature waterflood with substantial
remaining enhanced oil recovery (EOR) potential. Zones 2
and 3 have been essentially water swept. Zone 4 is the primary
target for the flood today with the majority of the unswept oil
in the Zone 4B interval (upper half of Zone 4). Repeat Formation Testing (RFT) completed in 1994 on several
wells showed that Zone 4 pressures were systematically lower
than Zone 3 pressures indicating much lower vertical
Statement of the Problem
permeability than earlier conceived. Well M-34 (Figure 5)
Performance of the Northwest Fault Block Waterflood shows this pressure change of over 100 psi between Zone 4
In early 1996 an analysis was undertaken to ascertain why the and Zone 3.
NWFB area was underperforming compared to forecasts of the
area. Performance predictions, using decline curve analysis,
indicated that the recovery (40%-45%) would fall short of the
expected 55%. Comparisons with other in-field analog areas
WATERFLOOD MANAGEMENT: A CASE STUDY OF THE NORTHWEST FAULT BLOCK AREA OF
SPE 63152 PRUDHOE BAY, ALASKA, USING STREAMLINE SIMULATION AND TRADITIONAL WATERFLOOD ANALYSIS 3
Figure 5: M-34 RFT taken 3Q 1994 the 200° F reservoir by seawater at 80° F will cause a
thermoeleasic stress reduction of more than 1000 psi (Ref. 4)
M-34 RFT The NWFB injection began with clean 80° F seawater,
then changed to 160° F produced water with small amounts of
Formation Pressure oil and solids. The extent of thermoelastic fracturing in the
3350 3400 3450 3500 3550 3600 3650 3700
NWFB was not fully appreciated until a more holistic injection
8400
analysis was undertaken. Injection well analysis began with a
thorough understanding of each wells history including
8500 Sag
drilling, perforating and stimulations. Most injection wells
Shublik
8600 TZ4
were converted from production so this included the time
period when the wells were producing. The next step included
W a te r G ra d ie n t= 0 . 5 0 p s i/ft constructing a Hall Plot (Ref. 5) for each injection well (Figure
8700
Depth SS
1.4 MM
indicating zone was already
taking water
Injection Well Diagnosis 1.2 MM
Event: Reperf Date:Cum 5/24/1990 : 22762530 Phase 3:
1992
The waterflood strategy of the NWFB was to inject water to Event: Reperf Date:Cum 7/31/1990 : 22854328 Start
1991 Time: 4/24/1992 End Time: 9/25/1994
Start Cum Inj: 30000000 End Cum Inj: 44000000
match the allocated reservoir voidage of the predefined local 1. MM
Ave Rate: 20136 Ave WHP: 575 Ave II: 35
Cum WHP
1990
pattern area. This is the traditional method used throughout .8 MM
Phase 2:
the Prudhoe Bay Unit. Since the NWFB area was considered Start Time: 12/27/1987 End Time: 4/23/1992
Start Cum Inj: 1000000 End Cum Inj: 30000000
.6 MM
a single flood unit no attempts were made to allocate Ave Rate: 24465 Ave WHP: 1082 Ave II: 22.6
1989
production and injection on a vertical basis. .4 MM
Phase 1:
Flow meters or spinners were used to estimate the vertical .2 MM
Start Time: 8/27/1987 End Time: 12/26/1987
Note: Phase 1 was very
short. II increase from 13.1 to
injection distribution by Zone. Unfortunately, flow meters 1988
Start Cum Inj: 2570 End Cum Inj: 1000000
Ave Rate: 8061 Ave WHP: 622 Ave II: 13.1
22.6 indicates probable
fracturing.
only indicate near wellbore flow, which can be very . MM
. MM
1987
5. MM 10. MM 15. MM 20. MM 25. MM 30. MM 35. MM 40. MM
Streamline Flood Front Tracking Simulation Successful Streamline Simulation in the NWFB
Well Performance Expectations The streamline model is being used for forecasting, drilling,
After completing the decline, RFT, and injection well analysis well work, understanding injection well conformance, and
a 3-D streamline front tracking simulation model was dynamic allocations to manage the waterflood. Several
developed to integrate the understandings into one prediction. development drilling locations have now been successfully
By comparing streamline simulation results to well behavior drilled as per model predictions, additionally water shut-offs
and surveillance data, future expectations for each well can be have been implemented. New information is input and the
viewed with more confidence. entire simulation rerun in 1-2 hours. This allows decisions to
Fit for Purpose Simulation be made on a daily basis, an impossibility for conventional
Due to the highly complex nature of the waterflood areas in simulation.
Prudhoe Bay, engineers have had difficulty in predicting fluid Model cross sections in Figure 7, show the structural and fluid
movements. To accurately resolve structure, geology, and complexity being resolved. Each model cell is 400’ x 400’
fluids, models sizes start at 300,000 cells with 200-300 wells, and approximately 20’ - 30’ thick. Water saturation at each
covering 23 years of production history. High resolution time step shows flood fronts advancing from water injection
faulted models are key to predicting fluid movements in the and aquifer influx. Production from each well in the simulation
NWFB. is being matched, and flood fronts match open hole logs as
The choices available to forecast reservoir performance in development drilling continues.
the area were generally classified into two categories.
Numerical simulation and analytical. Analytical approaches Figure 7: Model Complexity and Flood Front Movements
such as Dykstra Parsons, Stiles and material balance
techniques fail to discriminate communication between layers
in the reservoir. Numerical simulation techniques were then
examined to select the most appropriate tool.
Finite difference techniques have long been the norm for
the industry and particularly Prudhoe Bay. Full field models
with sophisticated wellbore hydraulic routines and surface
network packages have been used extensively for large scale
problem solving. In more recent times, grid refinement
techniques have been employed within the large full field
simulator to examine a smaller portion of the field. Full field
models and partial field models have been developed for the
Prudhoe Bay Field and the NWFB Area. Although these
methods can provide meaningful results, they are slow,
cumbersome, and have not provided a detailed prediction for
each well to help manage the waterflood at the well level.
Streamline simulation techniques have grown from simple 2
dimensional, 2 phase methods without gravity to today’s
versions which are 3 dimensional and account for gravity.
Streamline simulators have traditionally been very successful
in tracking fluid movement in the reservoir. Since this was
one of the primary goals in the analysis, streamline simulation Well by Well Production History Match over 23 years
was investigated further. Frontsim is providing a detailed NWFB history match of oil
Frontsim, a streamline based front tracking method, has and water production over the entire 23 year history of
become the simulation of choice for the NWFB since it Prudhoe Bay, for each well. Reservoir voidage is input, oil
addressed the most relevant issues in the area. Development and water rates and saturation are predicted. Because of the
and application of complex 3D models using Frontsim started maturity of water movements in the NWFB, detail for each
in Alaska in 1996. Full scale 3-D models in Prudhoe Bay are producing well in the history match provides for an accurate
now being developed and history matched in only a few means to quantify volumes from aquifer and injection.
months instead of years with the same size finite difference Key parameters of the history match were vertical
simulator. This allows engineers to focus on understanding permeability and injection well conformance. An iterative
reservoir and well behavior and not simply simulator process of matching flood fronts, initial breakthrough, and
development The streamlines directly quantify volumes water cut behavior was undertaken. Because of waterflood
between injectors and producers to provide a 3-D dynamic maturity, small changes in these key parameters made large
prediction of injection support. effects in the producers, indicating a relatively unique solution.
WATERFLOOD MANAGEMENT: A CASE STUDY OF THE NORTHWEST FAULT BLOCK AREA OF
SPE 63152 PRUDHOE BAY, ALASKA, USING STREAMLINE SIMULATION AND TRADITIONAL WATERFLOOD ANALYSIS 5
Several iterations were made each day and a timely solution Figure 8: Production history match Well S-08
reached.
The history match process was accomplished in stages.
The first stage was to match early water breakthrough response
from the aquifer before waterflood start. This was critical to
obtain the proper distribution of aquifer influx. Previous
material balance work had indicated that about 40,000 bwpd
was entering the NWFB Area from the aquifer. The aquifer
source was located where the Heavy Oil Tar(HOT) occurred in
Zone 3 conglomerates. The finely gridded model provided
detail in locating the HOT intersection with Zone 3. By
adjusting the relative volumes in each region a good match
was obtained.
The next stage involved regional flood front behavior. A
first pass at injection well conformance and permeability of the
open frame work conglomerates(within Zone 3) helped to
match flood fronts.
The next stage examined early water breakthrough timing
and watercut response in close wells. This was accomplished
by adjusting injection conformance and vertical permeability.
All permeability adjustments were on a layer basis over the
entire area.
Finally, adjustments were made to well completions..
These adjustments accounted for thermoelastic fracture growth
out of Zone 4 and into Zone 3 and 2 for the injectors, and
hydraulic prop-fraced in the producers. Usually this was
accomplished by opening layers below the perforated interval.
This adjustment provided virtually the 90% solution and the Solution to the Problem
most dominant portion of the history match process. Each of
these history match phases required an integrated Management of Injection Wells
understanding of the geology, wells, and displacement Understanding the vertical conformance of injected fluids was
process. the key to solving the under performing NWFB waterflood.
Figure 8 shows Well S-08 history matched rates and This single point can not be over emphasized.
cumulative production for oil and water predictions.
Historical GOR is presented for mechanistic review only(it Re-drilling Injection Wells
was not a history match parameter). Reservoir voidage and From March of 1997 to the present 16 injection wells have
injection are input. Output is oil and water rates along with been re-drilled in the NWFB region. These wells have
saturation for flood fronts for each time step. This is a typical confirmed the injection well analysis with open hole logs
match for each well in the model. Notice the large variations indicating 25 to 75 feet of oil column remaining in locations
in watercut resulting from well work and stimulation. This just 500 feet away from the original injection well. The well
type of detail is extremely difficult to match without accurate design utilized by these wells varied with the structure and
detail of flood front progression in three dimensions. stratigraphy expected, but generally high angle wells were
Streamline simulation provides a detailed history match of utilized. This well type was used to place injection into the
oil and water production over the entire 23 year history of unswept upper Zone 4 at high rates and at several injection
Prudhoe Bay, for each well. Because injection conformance is points along the well bore in hopes to contain the fracture
uncertain, due to near wellbore effects in the NWFB, model system and thereby increase vertical sweep conformance.
history match results confirmed injection well analysis that
indicated near wellbore effects were dominating injection well Setting Injection Targets with Dynamic Allocations
conformance. The history matched Frontsim model provided a means to
quantify the amount of production that could be supported by
injection. Because the NWFB shares pore volume with the
gravity drainage area, waterflood injection and production can
receive flux from the gas cap area. Figure 9 streamlines show
the ten year old fully developed NWFB waterflood. Static
6 G. H. GRINESTAFF, D. C CAFFREY SPE 63152
Decline Rate
11% per Year
10
Apr-77
Apr-79
Apr-81
Apr-83
Apr-85
Apr-87
Apr-89
Apr-91
Apr-93
Apr-95
Apr-97
Apr-99
Apr-01
Apr-03
Injection Targets using Frontsim Streamline Simulation
Injection well rates were set using the history matched
Frontsim model (Ref. 6). Model dynamic allocations called
Figure 10 shows streamlines over the same waterflood area for reducing the water injection rate from over 200 mbpd to
after injection well re-drills and injection allocation using 100 mbpd. This additional injection was not supporting
Frontsim was implemented. Injection at this point has been production, yet technical evaluations had not identified or
reduced by 40%. Flux out of the water flood area has been quantified the problem. The Frontsim simulation work was
eliminated, and flux between patterns (pattern skew) is implemented in late 1997 early 1998 and the resultant water
minimal. Flux from the gas cap is also starting to be mitigated production immediately dropped 50-60 mbd. Multiple effects
from re-drilling injection wells to increase support in upper were taking place on the oil stream at the time, but the
Zone 4. consensus is that oil rate has actually increased due to
hydraulic effects and we are currently experiencing and incline
Figure 10: Dynamic Allocations 1998 NWFB Prudhoe Bay in oil production. This demonstrated by the monthly
production data in Figure 12.
400
Water Production NWFB Area Production
350 Oil Production
Water Injection
300
Liquid Rates mbbl/day
250
200
150
100
50
0
Apr-77
Apr-79
Apr-81
Apr-83
Apr-85
Apr-87
Apr-89
Apr-91
Apr-93
Apr-95
Apr-97
Apr-99
Apr-01
WATERFLOOD MANAGEMENT: A CASE STUDY OF THE NORTHWEST FAULT BLOCK AREA OF
SPE 63152 PRUDHOE BAY, ALASKA, USING STREAMLINE SIMULATION AND TRADITIONAL WATERFLOOD ANALYSIS 7