1997 Forum Series

1
SPE DISTINGUISHED LECTURER SERIES is funded
principally through a grant of the SPE
FOUNDATION The Society gratefully
acknowledges those companies that support the
program by allowing their professionals to
participate as Lecturers. And special thanks to
The American Institute of Mining,
Metallurgical, and Petroleum Engineers (AIME) for
their contribution to the program.
2
Integrated Reservoir ModelingChallenges and
Solutions

• Mohan Kelkar
• The University of Tulsa

3
Outline

• Background
• Approach
• Hierarchical Descriptions
• Dynamic Data Integration
• Ranking
• Upscaling
• History matching of multiple descriptions
• Uncertainty representation
• Future Challenges
• Conclusions

4
Background

• What is integrated reservoir modeling?
• Integration of various qualities and
  quantities of data to generate inter well
  reservoir properties of interest so that
  uncertainties in future reservoir performance can
  be predicted.

5
Background

• What are the challenges?
• Scale and resolution of input and output
• Size of geomodel vs. size of simulation model
• Quantification of Uncertainties
• Solutions of inverse problem, especially during
  history matching of production data

6
Background

• Drawbacks related to Conventional History
  Matching
• Geological and geophysical uncertainties
• Uncertainties in future performance.
• The relationship between scale and uncertainty.

• Drawbacks related to Automatic History Matching
• Computationally intensive
• Customization to appropriate input parameters
• Objective Function
• Initial Guess Dependent

7
ApproachWork Flow
Structural Modeling
Hierarchical Realizations Property
Modeling Seismic Porosity Integration Limited
Dynamic Data Integration Fluid in Place
Calculations
Well Logs Generation (Rock Type, Perm)
Spatial Modeling
3 Selected Realizations
Selective History Matching
Upscaling Of Prop.
Ranking of Realizations
8
Topics of Concentration

• Hierarchical Descriptions
• Well Test Integration
• Upscaling using dynamic characteristics
• Objective history matching
• Future uncertainties representation

9
Hierarchical Multiple Realizations

• Rank the uncertain parameters from the largest to
  the smallest scale
• Discretize the range of uncertainties if possible
• Use fewer number of realizations for small scale
  uncertainties
• Limit the potential number of realizations to
  less than hundred
• Use methods such as experimental design to
  efficiently sample the range of uncertainties in
  input parameters

10
Limited Dynamic Data Integration

• Well Test Data
• Adjustment of fine scale permeabilities through
  adjustment factors accounting for fractures,
  multi-phase and scale
• PLT (Production Log Testing) Data
• Vertical adjustment to account for flow
• Determination of fracture conductivity

11
Matching PermeabilityProcedure
Simulated Fine Scale Permeability Distribution
Well Test
Re
Radial Upscaling
Stop
Yes
KH - Well Test
Match ?
KH Sim
No
Enhanced Permeability
Fracture
12
Alteration without Fractures

• Calculate the upscaled value of kh from fine
  scale description
• Calculate the ratio of (kh)well test to
  (kh)upscale.
• Interpolate the ratio across the field using
  kriging or similar technique
• Adjust the fine scale permeability value
  accordingly

13
Matching PermeabilityBackground Enhancement

• Definition
• Enhancement required to match well test when
  there is no fracture.
• Physical Interpretation
• Enhancement required due to micro
  fracture/fissures which are not captured by
  seismic curvature analysis

14
Matching PermeabilityLog (EF) vs Fracture Density
Background Enhancement
15
Permeability before and after enhancement
Layer 35 Not-enhanced
Layer 35 Enhanced
16
Permeability before and after enhancement
Layer 35 Enhanced
Layer 35 Enhanced
17
Permeability Anisotropy
18
Permeability Anisotropy

• Assume that permeability in the direction of
  fractures is maximum permeability and the one
  perpendicular to that is the minimum
  permeability. Minimum value is the base value.
• The enhanced permeability is calculated as
• Based on tensor relationship

• Assume that permeability in the direction of
  fractures is maximum permeability and the one
  perpendicular to that is the minimum
  permeability. Minimum value is the base value.
• The enhanced permeability is calculated as
• Based on tensor relationship

• Assume that permeability in the direction of
  fractures is maximum permeability and the one
  perpendicular to that is the minimum
  permeability. Minimum value is the base value.
• The enhanced permeability is calculated as
• Based on tensor relationship

• Assume that permeability in the direction of
  fractures is maximum permeability and the one
  perpendicular to that is the minimum
  permeability. Minimum value is the base value.
• The enhanced permeability is calculated as
• Based on tensor relationship

19
Dynamic Ranking

• Use the information from all the realizations
• Use different methods to rank realizations
• Permeability connectivity
• Streamline simulation
• Finite difference simplified simulation
• Use observed parameter of interest
• Select three to five realizations for history
  matching

20
Dynamic Ranking
21
UpscalingVertical Upscaling Optimization -
Procedure
Fine Scaled Model
22
Upscaling Scenarios

• Coarsen the geo-cellular grids while preserving
  the necessary level of heterogeneity
• Use streamline simulator to calculate the sweep
  efficiency of each vertical layer
• Combine vertical layers having similar
  displacements
• Test the vertical upscaling scenarios with
  Streamline simulator
• Sweep efficiency of each vertical layer of the
  upscaled model should be close to the sweep
  efficiency of the fine scale model.
• Fine tune the scenarios if needed

23
Optimum Upscaling Level
Fine Scale (93 layers)
Coarse Scale (66 layers)
24
Optimum Upscaling Level
Fine Scale (93 layers)
Coarse Scale (50 layers)
25
Optimum Upscaling Level
Fine Scale (93 layers)
Coarse Scale (30 layers)
26
Optimum Upscaling Level
Fine Scale (93 layers)
Coarse Scale (20 layers)
27
Upscaling Optimization
28
Rock Type Upscaling
246 Layers
57 Layers
29
Porosity Upscaling
246 Layers
57 Layers
30
Permeability Upscaling
246 Layers
57 Layers (Kx)
31
Compare Well Logs Before and After Upscaling
32
History Matching

• Define objective standards for history matching
• Vary dynamic parameters within the range of
  uncertainty
• Explore the impact of input parameters on
  observed performance
• Simultaneously history match multiple realizations

33
Field Example 1

• Carbonate, naturally fractured reservoir
• Influence of water influx as well as injected
  water
• Approximately 55 well strings
• Parameters adjusted
• Relative permeability parameters
• Aquifer strength
• Local permeabilities at three wells

34
Field StudyHistory Match
95
59
38
History Matching
Well Testing
Static Model
Stage 2 Calibrated Model
Stage 1
Stage 3 Final Model
Original Model
35
Stage-1 (38)
Field-wide Match
Stage-3 (95)
Stage-2 (59)
36
Stage-1 (38)
37
Stage-2 (59)
38
Stage-3 (95)
39
Field StudyHistory Match (contd)Middle Zone
Matrix Well
40
Field StudyHistory Match (contd)Upper Zone
Fracture Well
BHCIP/PBU
BHFP
Oil Rate
Water Cut
41
Field StudyHistory Match (contd)

• Blind Tests
• 7 Newly Drilled Wells
• 3 Rehorizontalized Wells
• 4 New Horizontal/High Deviated Wells
• 17 Pressure Observer Well Strings
• 10-months Extended Production Data
• Results
• 6 out of 7 well production was successfully
  simulated
• 15 out 17 pressure observation wells were matched
  
• Excellent Field-wide performance during the
  extended period

42
Field StudyBlind Test at Rehorizontalized Well
BHCIP/PBU
BHP
Oil Rate
Water Cut
43
Field StudyBlind Test at New Fractured Well
BHCIP/PBU
BHFP
Water Cut
Oil Rate
44
Field StudySaturation Comparison at New Wells
45
Field StudyPressure Match at Observer Well
BHCIP
46
Field StudyBlind Test at the Extended Production
Time Period
Reservoir Pressure
Cum. Oil
Water Cut
Oil Rate
47
Field Example 2

• Highly faulted sandstone reservoir (over 100
  faults)
• Large uncertainty with respect to permeability
  values
• More than 110 wells producers and injectors
• Weak aquifer drive high water cut in many wells
• Parameters adjusted
• Aquifer strength
• Relative permeability parameters
• Capillary pressure curves
• Fault transmissibilities

48
History Matching Field Level
FLPR (stb/day)
Time, days
FGOR (mscf/stb)
Time, days
49
History Matching Field Level
FOPR (stb/day)
Time, days
FWCT
Time, days
50
History Matching Group Level
Bloque4
GGPR mmscf/day
Time, days
Bloque5
GGPR mmscf/day
Time, days
51
History Matching Group Level
Bloque4
GOPR mstb/day
Time, days
Bloque5
GOPR mstb/day
Time, days
52
History Matching Group Level
12
Bloque4
8
GWPR mstb/day
4
0
Time, days
Bloque5
8
GWPR mstb/day
4
0
Time, days
53
History Matching Well Level
W31
400
WOPR stb/day
200
0
Time, days
W31
800
WWPR stb/day
400
0
Time, days
54
History Matching Well Level
2000
W90
1600
1200
WOPR stb/day
800
400
0
Time, days
3000
W90
2000
WWPR stb/day
1000
0
Time, days
55
History Matching Well Level
1200
W25
800
WOPR stb/day
400
0
Time, days
3000
W25
2000
WWPR stb/day
1000
0
Time, days
56
Future Challenges

• Right Scaling of Reservoir Model
• Generate reservoir description consistent with
  resolution of production data
• Generate reservoir description consistent with
  the flow process in the future
• Prioritize Observations
• Some observations are more important than others
• Large perturbations have more information content
  than small perturbations
• Eliminating large amount of observations prior to
  history matching will make the process cleaner
  and easier

57
Future Challenges

• Uncertainty Quantification in Future Performance
• Fit for purpose uncertainty quantification
• Quantification during the exploration phase
• Use of uncertainties prediction in future
  reservoir management

58
Conclusions

• A practical work flow allows an efficient history
  matching of multiple reservoir descriptions
• Partial integration of dynamic data makes the
  history matching more efficient
• Uncertainties in future performance can be
  quantified through multiple reservoir
  descriptions

59
Maintain Local Consistency among Attributes