Transcript Erosion in Surface-based Modeling Using Tank Experiment Data

Erosion in Surface-based Modeling
Using Tank Experiment Data
Siyao Xu, Andre Jung,
Tapan Mukerji, Jef Caers
SCRF 2012
Erosion in Reservoirs: Removed Flow
Barriers
• Erosion on fine-grained layers in a lobate
reservoir significantly affect reservoir
performance
Subseismic scale fine-grained layers
Eroded fine-grained layer by following events
SCRF 2012
After Alpak et. al. 2010
Erosion in Surface-based Modeling:
Uncertainty of Geometry
• Objective:
– Characterize uncertainty of erosion geometry
– Simulate erosion in surface-based model
Subseismic scale fine-grained layers
Eroded fine-grained layer by following events
SCRF 2012
After Alpak et. al. 2010
Review of Surface-based Modeling:
Advantages
• Can represent complex geometry
• Realistic model with low computational costs
Statistics & rules
from data and
knowledge
Sequentially generate
geometry with statistics
and place with rules
Stack Surfaces
to construct 3D
model
T
After Bertoncello et. al. 2011
SCRF 2012
Review of Surface-based Modeling:
Challenges
• Limited knowledge of erosion geometry
• Where to get data?
Statistics & rules
from data and
knowledge
Sequentially generate
geometry with statistics
and place with rules
Stack Surfaces
to construct 3D
model
T
After Bertoncello et. al. 2011
SCRF 2012
Tank Experiment Data
• Experiment data for reservoir modeling
– Detailed information of the processes
– How to use it to build a reservoir model?
• Distributary delta experiment
– Chris Paola, National Center for Earth-surface
Dynamics(NCED)/St. Anthony Falls Laboratory
(SANL)
SCRF 2012
Tank Experiment Data: Settings
5m
5m
Infeed point
2.5 m
Tank
SCRF 2012
Tank Experiment Data: Two Types of Data
• 1. Overhead photos taken at intervals of 15 s
T = 15 s
T=0s
T = 30 s
SCRF 2012
Tank Experiment Data: Two Types of Data
• 2. Elevation lines measured at 2 min interval
– Measured along three lines
– Distal : 2 m to the infeed point
Infeed point
2m
SCRF 2012
Tank Experiment Data: Two Types of Data
• 2. Elevation lines measured at 2 min interval
– Measured along three lines
– Medial : 1.75 m to the infeed point
Infeed point
1.75 m
SCRF 2012
Tank Experiment Data: Two Types of Data
• 2. Elevation lines measured at 2 min interval
– Measured along three lines
– Proximal : 1.5 m to the infeed point
Infeed point
1.5 m
SCRF 2012
Tank Experiment Data: from Overhead
Photos
Horizontal geometry
– Channels: belts of constant width
– Lobes: projecting oval shape
SCRF 2012
Tank Experiment Data: from Elevation Lines
Vertical geometry
– Patterns of erosion - deposition
Old Elevation
New Elevation
SCRF 2012
Tank Experiment Data: from Elevation Lines
Vertical direction
– Patterns of erosion - deposition
Old Elevation
No event
New Elevation
Erosion: New Elevation
Is LOWER than old elevation
Deposition: New Elevation
Is HIGHER than old elevation
SCRF 2012
Elevation
Tank Experiment Data: from Elevation Lines
SCRF 2012
Time Sequence
Elevation
Tank Experiment Data: Visualize Geometry
Horizontal
Location is
maintained
Height of geometry
is maintained
SCRF 2012
Elevation
Tank Experiment Data: Visualize Geometry
Time Sequence
Significant Geometry is maintained
New geometry is above previous
representing temporal sequence
SCRF 2012
Elevation
Tank Experiment Data: Visualize Geometry
Time Sequence
A larger deposition
occurs subsequently
SCRF 2012
Tank Experiment Data: Visualize Geometry
• Correlated
− erosion –
deposition
geometry
Correlated
geometry pairs
• How to
characterize the
statistics?
SCRF 2012
Tank Experiment Data: Interpretation and
Characterization
1. From channels: erosion dominated
WDeposition
WErosion  WDeposition
WErosion
SCRF 2012
Tank Experiment Data: Interpretation and
Characterization
1. From channels: characterize with DIMENSIONLESS
ratio
hDeposition
hErosion
rChannel
hErosion

hDeposition
SCRF 2012
Tank Experiment Data: Interpretation and
Characterization
2. From lobes: deposition dominated but with light
erosion
WDeposition
WErosion
WErosion  WDeposition
SCRF 2012
Tank Experiment Data: Interpretation and
Characterization
2. From lobes: characterize with DIMENSIONLESS
ratio
hDeposition
hErosi on
rLobe
hErosion

hDeposition
SCRF 2012
Tank Experiment Data: Interpretation and
Characterization
3. From lobes: deposition without erosion
characterized with probability
f 
N Lobes with Erosion
N Lobes
 0.29
SCRF 2012
Model Building: Generate Geobody
• Generate boundary along a centerline
Centerline Controlled Channel-Lobe Prototype Boundary
• Centerline is controlled by multiple
geological meaningful points
Channel Source
Thickest Point
Channel Intermediate Points Channel End
Boundary Control Point
SCRF 2012
Lobe End
Model Building: Generate Geobody
hDeposition
hErosion
hDeposition
hErosion
SCRF 2012
Model Building: Place Geobody
Distance to source
Combined Probability Map
1
Distance to previous lobe
Combined
0
Deposited thicknes
Next lobe location is picked-up
using the combined P map
SCRF 2012
Simulation Results: Reproduction of
Erosion-Deposition Processes
Section 1
Section 2
SCRF 2012
Simulation Results: Reproduction of
Erosion-Deposition Processes
Section 1
Section 2
SCRF 2012
Simulation Results: Reproduction of
Erosion-Deposition Processes
Section 1
Section 2
SCRF 2012
Simulation Results: Reproduction of
Erosion-Deposition Processes
Section 1
Section 2
SCRF 2012
Simulation Results: Reproduction of
Erosion-Deposition Processes
Section 1
Section 2
SCRF 2012
Simulation Results: Reproduction of
Erosion-Deposition Processes
Section 1
Section 2
SCRF 2012
Simulation Results: Reproduction of
Erosion-Deposition Processes
Section 1
Section 2
SCRF 2012
Simulation Results: Reproduction of
Erosion-Deposition Processes
Section 1
Section 2
SCRF 2012
Simulation Results: Complex Surfaces
Realization of 30 events
SCRF 2012
Simulation Results: Uncertainty of Surfaces
Cross sections orthogonal
to the flow direction
Colors Identify Different Events
SCRF 2012
Simulation Results: Uncertainty of Surfaces
Cross sections along the
flow direction
Flow Direction
Colors Identify Different Events
SCRF 2012
Simulation Results: Reproduction of Input
Statistics
• The probability of lobe with erosion is
reproduced f = 0.32
• Correlated erosion – deposition are reproduced
Experiment
Simulation
SCRF 2012
Simulation Results: Reproduction of Input
Statistics
Q-Q plots
rChannel
SCRF 2012
Simulation Results: Reproduction of Input
Statistics
Q-Q plots
rLobe
SCRF 2012
Conclusions and Future Works
• Conclusions
– Geometry information of erosion – deposition
is extracted from tank experiment data
– Realization of surface-based model can
reproduce input statistics from tank
experiment
• Future works
– How to combined horizontal and vertical
information?
– How to extract information to improve the pmaps
SCRF 2012