FD

Shell Exploration & Production
Integrated Waterflood Course
Pattern Selection & Well Spacing

Pattern Selection - Impacts
• Sweep efficiency
• Injectivity
• Flexibility to modify
• Regulators
5-Spot Pattern 7-Spot Pattern

Producer-to-Injector Ratio = 1:1 Producer-to-Injector Ratio = 2:1
9-Spot Pattern Linedrive Pattern

Producer-to-Injector Ratio = 3:1 Producer-to-Injector Ratio = 1:1
2 Technical & Operational Excellence 22 August 2018

Pattern Selection – Peripheral
• Injectors placed to supplement aquifer

– consider rock property variations near OWC
• Best when vertical communication and/or dip is high (gravity
stable)
• Common for Shell offshore Gulf of Mexico
• Aquifer influx models
– Schilthuis - 1936 (steady-state)
– van Everdingen & Hurst - 1949 (un-steady-state)
– Carter-Tracy - 1960 (un-steady-state)
– Fetkovitch – 1971 (pseudo-steady-state)
• neglects the transient period
• popular
• fairly accurate
– Coats, Allard & Chen for bottom-water drive
– see Chapter 8 Craft, Hawkins & Terry for example calculations
• Strongly consider using numerical simulation

Pattern Selection – Pattern
• Lack of natural water drive

• Low dip
• Implies some degree of control
• Simple analytical models are available
– various mobility ratios and well skins
– simplifying assumptions
– restricted on range of sensitivities
• Example of non-repeating patterns
– Milne Point Field

Patterns - Milne Point Field (Kuparuk A-Sand)

Pattern Selection – Pattern (continued)
• Ensure reasonable hydraulic connectivity between the injector

and producer
• Avoid short-circuits through fractures, thief zones and
conductive faults
• Consider injector rows along axis of maximum horizontal stress
– reduce short circuit via induced fractures
• Balance productivity & injectivity
• Additional considerations
– fault/fractures
– areal heterogeneity
– reservoir anisotropy
– mobility ratio (show streamlines)
– pattern conversion flexibility needs
• SPE 75140 Producer to Injector Ratio by Hansen

Well Spacing
• Economics!
• Hydraulic connectivity
• Permeability (effective)
• Anisotropy
• Stimulation techniques
– acid stimulations
– fracturing
• Well design & trajectory

Well Spacing
• Injection above the fracturing gradient

– increase water injectivity
– reduce the number of injectors
– monitor and control fracture growth
– reduced sweep efficiency?
– injection loss into non-target zones
– proppant & proppantless injector fractures
– thermal fracturing
– many waterfloods operate under fracture conditions
• Recovery mechanisms
• Phased development with infill drilling

Well Patterns, Spacing, & Sweep Efficiency
• Injector & producer

– shortest streamline
– highest pressure gradient
– @ breakthrough only a fraction of the area has been swept
– analytical solution for unit mobility
– physical experiment for non-unit mobility
– numerical simulation
• finite difference
• finite element (streamline simulation)
• generates recovery efficiency value
• show example

• Depleted for 6 years, then waterflood for 44 more

years
• Inverted 9-Spot (injector & 8-producers)
• Effect of areal heterogeneity

– Case 1 – porosity (12%) and permeability (100 md.)
– Case 2 – porosity and permeability vary randomly
• Water saturation plots

– red for maximum
– blue for minimum

20 Years & Constant Properties

50 Years & Constant Properties

Non-Constant Properties

20 Years & Non-Constant Properties

50 Years & Non-Constant Properties

• Simulation results quite idealized

• Heterogeneity in static properties still resulted in only
slight skew in saturations
• Well operations were fixed
– how realistic is this?
• History match model to account for differences
• Comfort in choosing new well locations?

Well Patterns – Additional Influences
• Existing well stock

– a sidetrack may also provide a additional capabilities
• Surface & subsurface topography
• Well types
– vertical, high slant, horizontal, designer
• Reservoir characteristics
– gas cap
• Boundary conditions
– subsurface (GDWFI)
– surface
• Influenced primary recovery performance
– detection of faults/fractures

Analytical Performance Predictions

Waterflooding – Analytical Methods
• Buckley-Leverett Model
– best suited for:
• small-scale applications
• moderately permeable
• relatively light oil
• reservoir layer is thinner than the capillary transition zone
• flow instability due to viscous fingering is not present
– key findings:
– Sorw impacts ultimate recovery
– relative permeability curves shapes are important
– oil-wet reservoir: more injection needed to achieve ultimate
recovery
– higher oil viscosity: higher the water cut at breakthrough and the
slower the oil recovery
• large amount of water recycling even in homogeneous
reservoir

• Dietz Model
– Given small capillary pressures (e.g., high permeability sandstone)
gravity segregates oil from injected water
– formula for calculating the tilt angle of the displacement front
relative to the bedding of the reservoir
– a function of:
• dimensionless gravity number
• dip-angle
• mobility ratio
– calculates the critical injection rate above which injected water will
under-run the oil and form a tongue
– key finding: displacement is stable in a tilted reservoir when the
mobility ratio is less than or equal to one

Waterflooding – Dietz Model

• Stiles and Dykstra-Parsons Models

– first generation analytical technique useful for forecasting
waterflood recovery in a layered reservoir
– piston-like displacement
– non-communicating layers.
– rarely used in their original forms
• recent adaptation replaced linear flow by radial flow in a 5-spot
– key findings:
• sensitive to permeability contrast between layers
– Dykstra-Parsons coefficient
– Lorentz coefficient

Waterflooding – Stiles and Dykstra-Parsons
Homogeneous Reservoir Heterogeneous Reservoir
Water Oil Water Oil
Low Dykstra -Parsons High Dykstra -Parsons

Coefficient Coefficient

Waterflooding – Buckley-Leverett with Gravity
• Gravity along the bedding plane is considered

• Capillary pressure gradients in this direction are
ignored
• Can be no saturation change over the height - the
length of the capillary transition zone greatly exceeds
the height of the formation
• Can be no "viscous fingering“
• Worked example in Shell Production Handbook Vol.
4

Waterflooding – Dietz w/ Saturation Transition
• Combines Dietz with Buckley-Leverett

• Calculates shape of the displacement front together
with the saturation transition behind the front
• Unstable displacements do not always occur even if
the endpoint mobility ratio is unfavorable
• When calculating conditions of stable displacement
and displacement front angles, the validity of simple
models such as Dietz should first be verified

Waterflooding – Stiles Model for 5-Spot Pattern
• Gardner’s technique replaces the piston-like displacement of

Stiles
• Radial frontal displacement in a pattern-flood situation
• Quick forecasts of oil recovery of a waterflood
• 5-spot pattern
• Uniform reservoir properties
• “Shell’s Waterflood Spreadsheet”
• Variables:
– Dykstra-Parsons coefficient
– water-oil relative permeabilities
– viscosities
– initial gas saturation
• Shell Report EP 93-2361

Water Injectivity
Water Injectivity
• Injectivity: injection rate per unit pressure drop

• Analogous to productivity index for a producer
• Function of:
– permeability
– thickness
– skin
– pattern geometry
• Assuming unit mobility ratio, injectivity is independent of:
– sweep efficiency
– time
• Assuming non-unit mobility ratios the injectivity changes as the
displacement progresses:
– improves if M>1
– degrades if M<1

Water Injectivity
• Theoretical change in injectivity M=4

Water Injectivity
• Analytical methods are over-simplified

• Severe limitations in their use
• Physical modelling would provide a more robust
solution
• Analytical & physical models assume constant rock
properties
• Fine-scale simulations
– more complexities
– variable properties
• History match field performance data

Integrated Reservoir Modeling

SB3
TST
SB2
HST
MFS
TST
SB1
HST
MFS
TST
Seismic Well logs
Analogue
Outcrop Image Log
Data
Core Drilling logs
Well Field
Test Production Analogue
Data

Seismic
• visualisation Deliverables
• seismic conditioning • maps
• inversion
• volumes
• reserves
•integrated 3D
reservoir model
Logs •drainage points
• correlation • reservoir
• properties
• inversion/
“bodies”
neural nets • field
Static 3D development
Models
• well planning
• volumes
• properties
Dynamic Models
• drainage points
• well forecasts
• linked to geology
• field development

It’s all a matter of scale !!

Tool
modelling X-well Seismic
tomography
Core analysis
Logs Seismic
Pore Borehole
modelling geophysics
Logs
Cores
SEM
mm cm m 10-100m > 100m
FLOW UNITS
RESERVOIR COMPARTMENTS
FIELD SCALE
Km’s

Reservor Characterisation –
Pore Systems
Coarse bioclastic
grainstone with large
connected vugs C Ma-5 core (3615-3685 m Ahd)
1000
100
Coral pebble with
K md(air)
3615-3624 m
low Phi ( intraskeletal
pores) and large B 10
3627-3632 m
solution vugs
3639-3642 m
1
3654-3683 m
K = 10^(-2.45 + 17*Phi)
Fine Bioclastic 0.1
wackestone with A 0.00 10.00 20.00 30.00 40.00
small vugs Phi (%bv)

Are there facies associated cements (e.g. flooding surface

carbonates) or clays (e.g. low resistivity smectites) ?
Is there a relationship with depth or structure ?
Ф/K relationships
Pore-bridging illite
(high Ф low K)
Grain coating
smectite
(low resistivity)
Polygenic
sandstone

Detailed Lithofacies and Correlation

Reservoir Properties from Seismic
Porosity from Sparse Spike

Inversion
Porosity %

Fracture Prediction Studies
Low mud loss
Fault
High mud loss

kmax
Volume of rock
fractured prior to
through-going
fault slip
Damage zone
(c. a few feet is
kmin usually low in

permeability.
Effective
process zone (a
few hundred feet) is
Background
fractures the zone that is
due to significantly
normal greater than
stress state. matrix flow.
Not to scale
After Ron A. Nelson,
42 2002 Excellence
Technical & Operational 22 August 2018
FACIES AND FRACTURE INTEGRATION
Sequence
Stratigraphy Fracture
Units Are Used Zone 48
to Constrain
Some of the
Fracturing.
Fault Seen in High Perm Corridors

Vicinity of W ell in Fracture Zones
35 35&48 to E xplain
Gas Breakthrough in
W ells Completed
Fracture Zone 35 Stratigraphically Low
W ATERFLOODING DID NOT W ORK


Challenges of
modeling…
Isolated
channels?
Major
isolated
channel
displaying
characteri
stics of
Or alternatively
lateral
accretion,
Huesca,
Spain.
Thin
correlatable
sheets?

Using Reservoir Simulations to Explore

Sensitivities of WF Forecasts to Many Uncertainties
Sensitivity Runs for WFDevelopment with300-Meter Infill Sensitivity of Forecasted Incremental Recovery
Al HuwaisahMainArea WF & 300-Meter Infill
Al Huwaisah Main Area
7000
6
6000
5
Cum Oil (Million M3)

5000 4
4000 3
Oil Rate(M3perDay)
3000 2
1
2000
0
1000
se
So x 10
c
q
So 40%
0%
FW m
Kv 5m
Kv 0
10 if
FW Aq
ra
xA
.1
u
5
Ba
Aq
_F
=2
L+
x0
L-
t
=
Bo
t_
Hi
rw
rw
0
o
_B
,N
Feb-70 Jan-75 Jan-80 Jan-85 Jan-90 Jan-95 Jan-00 Jan-05 Jan-10 Jan-15 Jan-20 No
Aq
nk
No_Bot_Aquif 10x Aq 10xFlankAq, NoBot Aq FWL-5m
la
FWL+5m Kv x 0.10 Kv x 10 Sorw=40% xF
10
Sorw=20% Hi_Frac Actual Base

1st hole “dry”

Discovery
A A A
Ultimate Recovery Estimates
D D D P15
Abandon
Uncertainty range
P85
Cumulative
Production
Prospect Appraise Devel. Proj. Devel.

Explore Reservoir Management
Evaluation Planning Exec. Drlg.
Time (not to scale)

1.0 Definition of Study Objectives
1.1 The objectives and scope of the study need to be evaluated and enumerated
before the study begins.
2.0 Data Collection, Validation and Interpretation
2.1 Evaluate available data
2.1.1 Data gaps
2.1.2 Discrepancies
2.2 Additional data requests
3.0 Determine “Fit-For-Purpose” Model Type & Design
3.1 Objectives
3.2 Timing
3.3 Costs
3.4 Manpower
3.5 Data
3.6 Future use (e.g., maintain for surveillance)
4.0 Model Construction
4.1 Geometry Data
4.1.1 Grid Selection – Grid Block Size & Orientation
4.1.1.1 Single cell
4.1.1.2 Radial
4.1.1.3 1D cross section
4.1.1.4 2D cross section/areal
4.1.1.5 3D
4.1.2 Grid Construction
4.1.2.1 Cartesian
4.1.2.2 Cylindrical
4.1.2.3 Curvilinear
4.1.3 Impact on run times and stability
4.2 Geological Model
4.2.1 1D, 2D and 3D Techniques
4.2.2 Model Types
4.2.2.1 Homogeneous
48 Technical & Operational
4.2.2.2 Excellence
Layer cake 22 August 2018
4.1.1.5 3D
4.1.2 Grid Construction
4.1.2.1 Cartesian
4.1.2.2 Cylindrical
4.1.2.3 Curvilinear
4.1.3 Impact on run times and stability
4.2 Geological Model
4.2.1 1D, 2D and 3D Techniques
4.2.2 Model Types
4.2.2.1 Homogeneous
4.2.2.2 Layer cake
4.2.2.3 Juxtaposed layers
4.2.2.4 Dual porosity
4.2.3 Reservoir Property Assignment by Averaging
4.2.3.1 Horizontal permeability
4.2.3.2 Vertical permeability and baffles
4.2.3.3 Pore volume and transmissibility
4.2.3.4 Modelling of faults and transmissibility barriers
4.2.4 Reservoir Property Assignment – 2D & 3D Sources
4.2.5 Compaction modelling
4.2.6 Aquifer modelling
4.2.7 Dual porosity behavior
4.3 PVT Data
4.3.1 Single phase
4.3.2 Black oil
4.3.3 Volatile oil & gas condensate
4.3.4 Compressibilities
4.3.5 Compositional
4.3.6 Chemical
4.3.7 Thermal
4.3.8 Variations with depth and area
4.4 Relative Permeability and Capillary Pressure Data
4.4.1 Rock relative permeabilities and capillary pressures
4.4.1.1 Evaluate by rock type
4.4.2 Pseudo relative Permeability and capillary pressures
4.4.2.1 Vertical equilibrium
4.4.2.2 Dynamic
4.4.3 Establish initial pressure and saturation distributions
4.4.1 Rock relative permeabilities and capillary pressures
4.4.1.1 Evaluate by rock type
4.4.2 Pseudo relative Permeability and capillary pressures
Integrated Reservoir Modeling 4.4.2.1 Vertical equilibrium
4.4.2.2 Dynamic
4.4.3 Establish initial pressure and saturation distributions
4.5 Well Data – Well Constraints and Hydraulics Models
4.5.1 Location
4.5.2 Constraints (pressure, rate, ratios, voidage)
4.5.3 Crossflow
4.5.4 Hydraulics
4.5.5 Dual porosity behavior
4.5.6 Single well
4.5.7 Multiple wells
4.5.8 Various well geometries (e.g., horizontal, multi-laterals)
5.0 Model Validation and Calibration
5.1 Material balance
5.2 Analytical methods
5.2.1 Displacement processes
5.3 History matching
5.4 Phases & steps
5.5 Criteria
5.5.1 Pressure
5.5.2 Ratios
5.5.3 Breakthrough timing
5.5.4 Contact movement
5.5.5 Productivity
5.5.6 Production logging
5.5.7 MDT’s/RFT’s
5.5.8 Pressure transient analysis
5.6 Reservoir description
5.7 Relative permeability
5.8 Fluid properties
50 5.9 TechnicalWell descriptions
& Operational Excellence 22 August 2018
5.10 Scenario manager
5.5.4 Contact movement
5.5.5 Productivity
5.5.6 Production logging
5.5.7
5.5.8
MDT’s/RFT’s
Pressure transient analysis
5.6 Reservoir description
5.7 Relative permeability
5.8 Fluid properties
5.9 Well descriptions
5.10 Scenario manager
5.11 Quality of match
5.12 Solution techniques & numerical methods
5.12.1 Stability
5.12.2 Grid block size/orientation/shape
5.12.3 Limiting numerical dispersion
5.12.4 Time steps
5.13 Regression analysis & automatic history matching
5.14 Upscaling
6.0 Forecasting
6.1 Well & reservoir parameters
6.1.1 Pressures
6.1.2 Rates
6.1.3 Ratios
6.1.4 Voidage
6.1.5 Saturation fronts
6.1.6 Recoveries
6.1.7 Well intervention requirements
6.2 Facility requirements
6.2.1 Artificial lift
6.2.2 Water handling
6.2.3 De-bottlenecking
7.0 Report
7.1 Objectives
7.2 Model type
7.3 Results/conclusions/recommendations
7.3.1 Maintaining model

Analytical Methods vs. Simulation
• Simulator is an excellent tool where there is a strong

dependence of the answer upon non-equilibrium,
generally time-dependent, spatial distributions of
pressure and fluid saturations.
• Such dependence makes element of symmetry or

analytical solutions unsatisfactory.

Waterflood Problem
Waterflood Problem
• Consider this idealized version of an offshore reservoir where at

least two injection wells will be needed to match expected
withdrawals. Because of platform constraints, no additional
wells can be drilled.
• Down dip limits of the aquifer are indefinite however, the aquifer
is believed to be relatively small and will not be able to support
the rates planned for development. The reservoir is just a few
hundred psi above the bubble point and water injection will be
needed from the beginning of production in order to maintain
rates and prevent the formation of a secondary gas cap up dip.

Waterflood Problem

Waterflood Problem
• Well 4 was planned to penetrate the reservoir

downdip from its actual location but just above the oil
water contact
• Well 5 was planned for a similar location down
structure from its present location.
• Wells 6 and 7 were intended for a different reservoir
and were so drilled.
• A subsequent restudy of the overall geology, original
correlations , etc., revealed that these two wells were
inadvertently completed in a different reservoir than
originally planned!

Waterflood Problem
• Here are four possible choices for conversions.
• Which one would you choose?

Waterflood Problem

Waterflood Problem
• A set of simulation runs were made with the four

choices shown.
• Overall the best choice was Case 4 both from an

ultimate recovery as well as present value return on
investment.
• This case minimized the volume of water to injected

as well as having the least resaturation losses
(pushing oil downdip into the aquifer) of any of the
patterns.

Waterflood Problem
• Case 1 pushed a lot of oil past the downdip

producers and into the aquifer.
• Case 2 showed a very rapid drop in production rate

as Well 7 quickly watered out Wells 5 and 6.
• Case 3 also showed a drop in production with the

early loss of Well 5. In addition, a substantial volume
of oil downdip of Well 4 was pushed towards the
aquifer and was not produced.

Waterflood Pilots
Waterflood Pilots
• The Golden Rule For Pilots: Locate the pilot in a representative

(i.e., similar geology, fluid saturations & saturation history)
portion of the field.
• Properly locating the pilot area takes up-front analysis
• What if?
– total net pay was similar, but it was distributed differently?
– the saturation history varied?.
• Limitations:
– The smaller the pilot, the greater the risk of locating it in a non-
representative portion of the reservoir.
– The smaller the well count, the greater the impacts of under-performing
well(s).
– Oil migration into and out of the pilot pattern may skew results; a similar
problem occurs with injected water.

Waterflood Pilots
• In one study, a single injection pilot (inverted 5-spot) ended up

sweeping twice the pilot area - leading to an overly optimistic
assessment of performance.
• In another study the voidage replacement ratio (VRR) was
purposely set at a low value, and not surprisingly, significant oil
migrated into the pattern area again giving overly optimistic
results.
• In spite of the two aforementioned studies, more frequently, pilot
oil recovery is typically lower than that obtained by full-field
flooding.
– improved full-field performance is often a result of a “learning curve”.
• Gas saturation at the start of flooding also influences the water
injection rate and the time to fill up.
– the higher the gas saturation at the start of waterflooding, the less likely that
oil will migrate outside the pattern.

Waterflood Pilots - Hints
• Increase the well count, and/or the number of patterns. Multiple

contiguous patterns can be used to minimize oil migration
losses.
– a value of information exercise would need to be performed
• The primary recovery mechanisms should be understood (e.g.,
aquifer presence & strength); primary performance predictions
should be made.
• Check the condition ratio (actual well productivity / "ideal"
skin=0, well productivity); higher is better for analysis purposes.
If the condition ratio is too low, you’ll likely get oil migration into
the pattern.
• Define the pilot objectives
o water injectivity (fairly simple to assess with a pilot)
o time to waterflood breakthrough (WFBT)
o recoverable oil & water injection/production requirements

• Consider infill drilling and pattern re-configuration

flexibility when choosing the pattern (e.g., 7-spot vs.
9-spot).
• Downsizing the well spacings to get earlier results.
• Keep your injector rates proportional on a phi-h
basis.
• A strong directional permeability trend (whether
structural or stratigraphic) should be accounted for.
• Evaluate peripheral (vs. pattern) drive if the dip angle
is significant.

• If you're in a dual-phi system, you must consider the

fracture/fault orientation/length/density/continuity.
• Plan a robust surveillance program (e.g., core, logs,
production/injection logs, PTA, observation wells,
etc.).
• Estimate the mobility ratio and its impact on the
producer to injector ratio.
• Inverted (injector-centered) pilot pattern flooding may
not provide a high-confidence estimate of full-field
recovery, but volumetric coverage at breakthrough,
and directional permeability data can usually be
obtained.

FD

Uploaded by

Document Information

Copyright

Available Formats

Share this document

Share or Embed Document

Sharing Options

Did you find this document useful?

Is this content inappropriate?

Copyright:

Available Formats

FD

Uploaded by

Copyright:

Available Formats

Shell Exploration & Production

Integrated Waterflood Course

Pattern Selection & Well Spacing

5-Spot Pattern 7-Spot Pattern

9-Spot Pattern Linedrive Pattern

2 Technical & Operational Excellence 22 August 2018

• Injectors placed to supplement aquifer

4 Technical & Operational Excellence 22 August 2018

• Lack of natural water drive

5 Technical & Operational Excellence 22 August 2018

6 Technical & Operational Excellence 22 August 2018

• Ensure reasonable hydraulic connectivity between the injector

7 Technical & Operational Excellence 22 August 2018

8 Technical & Operational Excellence 22 August 2018

• Injection above the fracturing gradient

9 Technical & Operational Excellence 22 August 2018

• Injector & producer

10 Technical & Operational Excellence 22 August 2018

• Depleted for 6 years, then waterflood for 44 more

• Inverted 9-Spot (injector & 8-producers)

• Effect of areal heterogeneity

• Water saturation plots

11 Technical & Operational Excellence 22 August 2018

12 Technical & Operational Excellence 22 August 2018

13 Technical & Operational Excellence 22 August 2018

14 Technical & Operational Excellence 22 August 2018

15 Technical & Operational Excellence 22 August 2018

16 Technical & Operational Excellence 22 August 2018

• Simulation results quite idealized

17 Technical & Operational Excellence 22 August 2018

• Existing well stock

18 Technical & Operational Excellence 22 August 2018

Integrated Waterflood Course

Analytical Performance Predictions

20 Technical & Operational Excellence 22 August 2018

21 Technical & Operational Excellence 22 August 2018

22 Technical & Operational Excellence 22 August 2018

• Stiles and Dykstra-Parsons Models

23 Technical & Operational Excellence 22 August 2018

Homogeneous Reservoir Heterogeneous Reservoir

Water Oil Water Oil

Low Dykstra -Parsons High Dykstra -Parsons

24 Technical & Operational Excellence 22 August 2018

• Gravity along the bedding plane is considered

25 Technical & Operational Excellence 22 August 2018

• Combines Dietz with Buckley-Leverett

26 Technical & Operational Excellence 22 August 2018

• Gardner’s technique replaces the piston-like displacement of

27 Technical & Operational Excellence 22 August 2018

Integrated Waterflood Course

• Injectivity: injection rate per unit pressure drop

29 Technical & Operational Excellence 22 August 2018

• Theoretical change in injectivity M=4

30 Technical & Operational Excellence 22 August 2018

• Analytical methods are over-simplified

31 Technical & Operational Excellence 22 August 2018

Integrated Waterflood Course

Integrated Reservoir Modeling

Seismic Well logs

33 Technical & Operational Excellence 22 August 2018

34 Technical & Operational Excellence 22 August 2018

It’s all a matter of scale !!

35 Technical & Operational Excellence 22 August 2018

36 Technical & Operational Excellence 22 August 2018