Static Model

Integrated Reservoir Study

DATA
MANAGEMENT Data Flows

RESERVOIR STATIC DYNAMIC

CHARACTERIZATION MODEL MODEL

Work Flows
 Static Model

PVT/Fluid
Saturations Basic Engineering
Model Results

Property Static
Reservoir
Distributions Reservoir
Flow Units
Model Model
 Static Model Design
Stratigraphic layering/coordinates

or

Multiple Realizations

Volumetrics / Mapping
Assess Connectivity
Scale-Up for Flow Simulation
 Cell-Based Modeling
• Cell-based,
Lithology
stochastic, 3-D
modeling of
lithology, porosity
and permeability in a
sequential order
Porosity
– Preserves major
heterogeneities
– Statistical control
is obtained from Permeability
cores, logs,
seismic and
outcrops
Object-Based Modeling

• Object-based,
stochastic, 3-D
modeling of well
defined morphologic
objects
• Pseudo genetically
simulate
depositional history
• Statistical control is
obtained from
cores, logs, seismic
and outcrops
 Static Mode GOAL

Develop and quantify geological

reservoir architecture, define
flow units and establish 3D
distribution of flow properties for
reservoir simulator input
 Static Reservoir Modeling
Develop Reservoir Architecture/
Flow Units

Deterministic Model Stochastic Model

Cells Objects Cells Objects

Geocellular Model
(millions of cells)

Upscaling
(thousands of cells)

Reservoir Simulator
}
Dynamic
Modeling
Geological Model versus Reality

• Anisotropy
- Directional Variation in Flow Properties

• Heterogeneity
- Spatial Variation in Reservoir Architecture/
Properties
Lateral Pay Discontinuities
Producing Injection
well well

Reservoir Sandstone
Trapped oil

Shale Barrier
Completion Interval Inconsistencies
Producing Injection
well well

Trapped oil -
lateral pay
discontinuities

Trapped
Oil - Completions
 Lateral Pay Discontinuities
• Effect of infill drilling
Producing Infill Injection
well well well
 Scales in Reservoir Characterization
300 m Relative Volume
14
Gigascopic 50 m Well Test 10

300 m
Reservoir Model 12
Megascopic Grid Cell 2 x 10
5m 150 m

2m
Wireline Log 7
1m
Interval 3 x 10
Macroscopic cm 2
Core Plug 5 x 10

mm - mm Geological
Microscopic 1
(modified from Hurst, 1993)
Thin Section
Key Architectural Heterogeneties
• Layercake
– High continuity baffles
– Pinch outs
– Permeability Profiles
• Jigsaw
– Thief zones
– Orientation of flow units
– Permeability contrasts
– Local baffles
• Labyrinth
– Orientation flow units
– Interconnectivity (N/G < .35)
– Dead end gravity traps
 Geological Architecture Will
Determine the Modeling Approach
• Layer-cake
– Deterministic layer boundaries Deterministic
– Variogram based Phi-K calculations

• Jigsaw
Deterministic
– Indicator simulation of lithotypes &
– Variogram based Phi-K calculations Probabilistic

• Labyrinth
– Object based modeling of sand bodies
Probabilistic
– Variogram based Phi-K calculations
 Reservoir Flow Units
STATIC MODEL

PVT/Fluid
Saturations Basic Engineering Results
Model PTA, PDA, MBA

Reservoir Property Static

Flow Distributions Reservoir
Units Model Model
 Definition - Flow Unit

A mappable portion of the total reservoir, within

which geological and petrophysical properties
that affect the flow of fluids are consistent and
predictably different from the properties of other
reservoir rock volumes.
(modified from Ebanks and Others, 1992)
Some Characteristics of Flow Units
A flow unit is a specific volume of a reservoir,
which is composed of one or more reservoir
quality lithologies and any nonreservoir quality
rock types within that same volume, as well as the
fluids they contain

A flow unit is correlative and mappable at the

interwell scale

A flow unit zonation is recognizable on wireline

logs

A flow unit may be in communication with other

flow units
 Comments on Flow Units
• Flow units are internally consistent but may
not necessarily be entirely homogeneous in
geological and petrophysical properties

• Flow units may contain more than one reservoir

quality lithology.

• Flow units may contain nonreservoir features,

such as shale and cemented layers.
 Geological and Petrophysical Data Used to Define Flow
Units

Core Pore Petrophysical Gamma Ray Flow

Core Lithofacies
Plugs Types Data Log Units
Capillary
f vs k Pressure

5
4

(modified from Ebanks and others, 1992)

 Deterministic Methods
• Traditional mathematical, engineering,
geologic, and other scientific methods
that produce unique solution to the
posed problem without involving methods
of random functions.

“God does not play dice”

A. Einstein
 Stochastic Model
a definition
• A set of simplified, equiprobable
realizations of the studied object
that honors available data
• Properties at unsampled locations
are estimated using stochastic
methods
•Methods of random functions
used for the solution of geological
and engineering problems and
statistical evaluation of results
Deterministic Modeling
 Mapping Layer Properties

Contours = (Net Thickness) / (Net Average Porosity)

Color = Thickness (feet)
 Length of Shale Barrier as a Function of
Depositional Environment
100

Marine
Percentage Longer

Deltaic
Barrier
50

Delta Fringe
and
Delta Plain
Coarse Distributary
Point Channel
Bar
0
0 500 1000 1500 2000
Length of Shale (Feet)
 Log Property Mapping
A Seismic Guided Porosity Map
Geostatistical Modeling
Methodology in Geostatistics
• Define area of interest.
- Scoping analysis

• Model spatial continuity

- Variogram modeling

• Estimate spatial distribution

of properties
– Kriging
– Conditional simulation
– Object modeling
 Spatial Continuity
• Exists in most earth science data sets
• On the contour map low values tend to
be near other low values and high value
tend to be near other high values
• Porosity and permeability fields are
never truly random
• Variogram quantifies spatial continuity
Kriging (Estimation) and
Simulation
• Kriging
– properties
– definition and
derivation
– example
• Cross validation
• Simulation
 Kriging (Estimation)
• Developed by D. Krige 1951
• Linear, unbiased, exact, minimizes error
variance
Problems in Application of Kriging
to Reservoir Modeling
• Underrepresents the variability

• Deterministic and cannot be used for

estimation of uncertainty

•The fields generated tend to be

more Gaussian (normal distribution)
than original
 Cross Validation
Purpose:
To evaluate accuracy of prediction

• Drop one point at a time,

• Estimate value at the dropped location
using kriging
• Compare true and estimated values
 Cross Validation
100

10
K Estimated

0.1

0.01
0.01 0.1 1 10 100
K True
 Cokriging
Problem:
• Petrophysical data (e.g. porosity,permeability)
is sampled sparsely (i.e. at wells)

• Seismic data (amplitude) is sampled densely

but does not directly measure desired
property (e.g. porosity or permeability)

A Solution:
• Cokriging correlates desired undersampled
reservoir property to widely sampled
parameter
 Cokriging
Well Data Example Using
Variogram Well and Seismic
Model Data

Cross-
Variogram Cokriging
Model
Cokriged Map

Varioram
Model
Seismic Data (modified from Wolf and others, 1994)
Stochastic Modeling

Gaussian Simulation
(cell-based model)
 Simulation
• Krigging
– smooth, good for detecting trends
– bad for building reservoir simulation model
• Simulation
– reproduces histogram, honors variability
– good for building reservoir simulation model
– allows to assess the uncertainty with multiple
realizations
 Stochastic Reservoir Modeling
Reality Model
Distribution of Rock/Fluid Properties Distribution of the Rock/Fluid Properties

single true distribution multiple stochastic models

Recovery Process Recovery Process

actual process implemented numerical model of process

Field Response Field Response
 Simulation Vs. Kriging
Simulation Kriging

Output  Multiple Equiprobable  One Deterministic

Realizations Model

Properties  Honors Wells  Honors Wells

 Honors Histogram  Honors Variogram
 Honors Variogram
Image  Noisy  Smooth away from
 Same variability wells
everywhere
Use  Flow Simulation  Mapping
 Uncertainty Calculation  Volumetrics
 Simulation:
Quantifying Porosity Uncertainty

Probability of Porosity Probability of Porosity

above 8% above 12%
 Sequential Gaussian
Simulation
• Transform the data to normal distribution
• Assign data values to closest grid nodes
(known points)
• Establish random path through all grid nodes
• Visit each grid node once
 Sequential Gaussian
Simulation
• Visit each grid node once
– find informed (by input and previously simulated)
cells
– construct conditional distribution by krigging (local
mean and variance define Gaussian distribution)
– draw a random number (norminv(rand(),m,s)
from histogram
• Check the results
 Stochastic Realization of Permeability in a
Cross Section Through Eolian Sandstone
Well Well Well

No Vertical Exaggeration
Individual Blocks are 5 Feet by 50 Feet
Permeability - Khmax

<0.1 md 0.1 - 0.5 md 0.5 - 2.5 md 2.5 - 15 md >20 md

(modified from Cox and Others, 1994)
 Stochastic Realization of Facies in a Cross
Section Through Eolian Sandstone

Well Well Well

No Vertical Exaggeration
Individual Blocks are 5 Feet by 50 Feet

Low-Angle High-Angle Mixed Grain

Wind Ripple Wind Ripple Flow
(modified from Cox and Others, 1994)
 Stochastic Modeling

Object Model
Boolean Simulation Example
 Boolean Simulation of Sand Channels

Conditioning Data

Sand
Shale

(modified from Srivastava, 1994)

 Boolean Simulation of Sand Channels
Honoring Well Data
Sand bodies randomly located
to coincide with sands in wells

(modified from Srivastava, 1994)

 Boolean Simulation of Sand Channels
Interwell Bodies
Random sand body conflicts with
well and must be dropped or moved

(modified from Srivastava, 1994)

 Boolean Simulation of Sand Channels
Final Realization
Sand bodies added until net-to-gross
ratio reaches desired target

(modified from Srivastava, 1994)