Semi-automatic history matching applying parameter ... · Semi-automatic history matching. applying...

211
Smart Fields, Stanford University Integrated Operations 2007 07 Semi-automatic history matching applying parameter estimation techniques October 3, 2007 D. Echeverr D. Echeverr í í a Ciaurri a Ciaurri Smart Fields Consortium Stanford University

Transcript of Semi-automatic history matching applying parameter ... · Semi-automatic history matching. applying...

Page 1: Semi-automatic history matching applying parameter ... · Semi-automatic history matching. applying . parameter estimation techniques. ... • History matching ... Workflow. 8. facies.

Smart Fields, Stanford UniversityIntegrated Operations 200707

Semi-automatic history matchingapplying

parameter estimation techniques

October 3, 2007

D. EcheverrD. Echeverríía Ciaurria Ciaurri

Smart Fields ConsortiumStanford University

Page 2: Semi-automatic history matching applying parameter ... · Semi-automatic history matching. applying . parameter estimation techniques. ... • History matching ... Workflow. 8. facies.

Smart Fields, Stanford UniversityIntegrated Operations 200707

Outline

O

• Smart Fields Consortium

• The History Matching Problem

• Challenges and Trends

• History Matching at Smart Fields

• Conclusions

Page 3: Semi-automatic history matching applying parameter ... · Semi-automatic history matching. applying . parameter estimation techniques. ... • History matching ... Workflow. 8. facies.

Smart Fields, Stanford UniversityIntegrated Operations 200707

Outline

O

• Smart Fields Consortium

• The History Matching Problem

• Challenges and Trends

• History Matching at Smart Fields

• Conclusions

Page 4: Semi-automatic history matching applying parameter ... · Semi-automatic history matching. applying . parameter estimation techniques. ... • History matching ... Workflow. 8. facies.

Smart Fields, Stanford UniversityIntegrated Operations 200707

Closing the Loop

1

Page 5: Semi-automatic history matching applying parameter ... · Semi-automatic history matching. applying . parameter estimation techniques. ... • History matching ... Workflow. 8. facies.

Smart Fields, Stanford UniversityIntegrated Operations 200707

Closing the Loop

1

Page 6: Semi-automatic history matching applying parameter ... · Semi-automatic history matching. applying . parameter estimation techniques. ... • History matching ... Workflow. 8. facies.

Smart Fields, Stanford UniversityIntegrated Operations 200707

Closing the Loop

1

Page 7: Semi-automatic history matching applying parameter ... · Semi-automatic history matching. applying . parameter estimation techniques. ... • History matching ... Workflow. 8. facies.

Smart Fields, Stanford UniversityIntegrated Operations 200707

Closing the Loop

1

Page 8: Semi-automatic history matching applying parameter ... · Semi-automatic history matching. applying . parameter estimation techniques. ... • History matching ... Workflow. 8. facies.

Smart Fields, Stanford UniversityIntegrated Operations 200707

Closing the Loop

1

Page 9: Semi-automatic history matching applying parameter ... · Semi-automatic history matching. applying . parameter estimation techniques. ... • History matching ... Workflow. 8. facies.

Smart Fields, Stanford UniversityIntegrated Operations 200707

Closing the Loop

1

Page 10: Semi-automatic history matching applying parameter ... · Semi-automatic history matching. applying . parameter estimation techniques. ... • History matching ... Workflow. 8. facies.

Smart Fields, Stanford UniversityIntegrated Operations 200707

Closing the Loop

1

Page 11: Semi-automatic history matching applying parameter ... · Semi-automatic history matching. applying . parameter estimation techniques. ... • History matching ... Workflow. 8. facies.

Smart Fields, Stanford UniversityIntegrated Operations 200707

Closing the Loop

1

Page 12: Semi-automatic history matching applying parameter ... · Semi-automatic history matching. applying . parameter estimation techniques. ... • History matching ... Workflow. 8. facies.

Smart Fields, Stanford UniversityIntegrated Operations 200707

Closing the Loop

1

Page 13: Semi-automatic history matching applying parameter ... · Semi-automatic history matching. applying . parameter estimation techniques. ... • History matching ... Workflow. 8. facies.

Smart Fields, Stanford UniversityIntegrated Operations 200707

Closing the Loop

Field Development Optimization

1

Page 14: Semi-automatic history matching applying parameter ... · Semi-automatic history matching. applying . parameter estimation techniques. ... • History matching ... Workflow. 8. facies.

Smart Fields, Stanford UniversityIntegrated Operations 200707

Closing the Loop

Field Development Optimization

1

Page 15: Semi-automatic history matching applying parameter ... · Semi-automatic history matching. applying . parameter estimation techniques. ... • History matching ... Workflow. 8. facies.

Smart Fields, Stanford UniversityIntegrated Operations 200707

Stanford Smart Fields

2

Page 16: Semi-automatic history matching applying parameter ... · Semi-automatic history matching. applying . parameter estimation techniques. ... • History matching ... Workflow. 8. facies.

Smart Fields, Stanford UniversityIntegrated Operations 200707

Stanford Smart Fields

• Stanford University

2

Page 17: Semi-automatic history matching applying parameter ... · Semi-automatic history matching. applying . parameter estimation techniques. ... • History matching ... Workflow. 8. facies.

Smart Fields, Stanford UniversityIntegrated Operations 200707

Stanford Smart Fields

• Stanford University

– Energy Resources Engineering

– Geophysics

– Management Sciences and Engineering

– CEES

– Geology, Electrical Engineering, …

2

Page 18: Semi-automatic history matching applying parameter ... · Semi-automatic history matching. applying . parameter estimation techniques. ... • History matching ... Workflow. 8. facies.

Smart Fields, Stanford UniversityIntegrated Operations 200707

Stanford Smart Fields

• Stanford University

• Current consortium members

2

Page 19: Semi-automatic history matching applying parameter ... · Semi-automatic history matching. applying . parameter estimation techniques. ... • History matching ... Workflow. 8. facies.

Smart Fields, Stanford UniversityIntegrated Operations 200707

Stanford Smart Fields

• Stanford University

• Current consortium members

– BP, Chevron, CMG, Conoco Phillips,

ExxonMobil, Landmark, NTNU, Petrobras,

Shell, Saudi Aramco, Statoil, Total

– in discussions with several others

2

Page 20: Semi-automatic history matching applying parameter ... · Semi-automatic history matching. applying . parameter estimation techniques. ... • History matching ... Workflow. 8. facies.

Smart Fields, Stanford UniversityIntegrated Operations 200707

Scope of Work

3

Page 21: Semi-automatic history matching applying parameter ... · Semi-automatic history matching. applying . parameter estimation techniques. ... • History matching ... Workflow. 8. facies.

Smart Fields, Stanford UniversityIntegrated Operations 200707

Scope of Work

• Optimization techniques

3

Page 22: Semi-automatic history matching applying parameter ... · Semi-automatic history matching. applying . parameter estimation techniques. ... • History matching ... Workflow. 8. facies.

Smart Fields, Stanford UniversityIntegrated Operations 200707

Scope of Work

Oilfield Review 2006

• Optimization techniques– well placement

– well type

– reservoir and production system operations

3

Page 23: Semi-automatic history matching applying parameter ... · Semi-automatic history matching. applying . parameter estimation techniques. ... • History matching ... Workflow. 8. facies.

Smart Fields, Stanford UniversityIntegrated Operations 200707

Scope of Work

• Optimization techniques

3

Page 24: Semi-automatic history matching applying parameter ... · Semi-automatic history matching. applying . parameter estimation techniques. ... • History matching ... Workflow. 8. facies.

Smart Fields, Stanford UniversityIntegrated Operations 200707

Scope of Work

• Optimization techniques

• Data filtering and integration

www.halliburton.com

3

Page 25: Semi-automatic history matching applying parameter ... · Semi-automatic history matching. applying . parameter estimation techniques. ... • History matching ... Workflow. 8. facies.

Smart Fields, Stanford UniversityIntegrated Operations 200707

Scope of Work

• Optimization techniques

• Data filtering and integration

3

Page 26: Semi-automatic history matching applying parameter ... · Semi-automatic history matching. applying . parameter estimation techniques. ... • History matching ... Workflow. 8. facies.

Smart Fields, Stanford UniversityIntegrated Operations 200707

Scope of Work

• Optimization techniques

• Data filtering and integration

• History matching

3

Page 27: Semi-automatic history matching applying parameter ... · Semi-automatic history matching. applying . parameter estimation techniques. ... • History matching ... Workflow. 8. facies.

Smart Fields, Stanford UniversityIntegrated Operations 200707

Scope of Work

• Optimization techniques

• Data filtering and integration

• History matching

• Fast reservoir modeling and proxies

3

Page 28: Semi-automatic history matching applying parameter ... · Semi-automatic history matching. applying . parameter estimation techniques. ... • History matching ... Workflow. 8. facies.

Smart Fields, Stanford UniversityIntegrated Operations 200707

Scope of Work

• Optimization techniques

• Data filtering and integration

• History matching

• Fast reservoir modeling and proxies

• Uncertainty propagation

3

Page 29: Semi-automatic history matching applying parameter ... · Semi-automatic history matching. applying . parameter estimation techniques. ... • History matching ... Workflow. 8. facies.

Smart Fields, Stanford UniversityIntegrated Operations 200707

Scope of Work

• Optimization techniques

• Data filtering and integration

• History matching

• Fast reservoir modeling and proxies

• Uncertainty propagation

• Decision making

3

Page 30: Semi-automatic history matching applying parameter ... · Semi-automatic history matching. applying . parameter estimation techniques. ... • History matching ... Workflow. 8. facies.

Smart Fields, Stanford UniversityIntegrated Operations 200707

Model Order Reduction

4

(from Cardoso et al., 2007)

Page 31: Semi-automatic history matching applying parameter ... · Semi-automatic history matching. applying . parameter estimation techniques. ... • History matching ... Workflow. 8. facies.

Smart Fields, Stanford UniversityIntegrated Operations 200707

Model Order Reduction

4

• PCA, original model size N = 800

Injec. 1 Injec. 2

Prod. 1 Prod. 2

(from Cardoso et al., 2007)

Page 32: Semi-automatic history matching applying parameter ... · Semi-automatic history matching. applying . parameter estimation techniques. ... • History matching ... Workflow. 8. facies.

Smart Fields, Stanford UniversityIntegrated Operations 200707

Model Order Reduction

4

• PCA, original model size N = 800

(from Cardoso et al., 2007)

Page 33: Semi-automatic history matching applying parameter ... · Semi-automatic history matching. applying . parameter estimation techniques. ... • History matching ... Workflow. 8. facies.

Smart Fields, Stanford UniversityIntegrated Operations 200707

Model Order Reduction

4

• PCA, original model size N = 800

full model N = 800n = 1 + 7 = 8n = 2 + 7 = 9n = 8 + 26 = 34n = 23 + 47 = 61

(from Cardoso et al., 2007)

Page 34: Semi-automatic history matching applying parameter ... · Semi-automatic history matching. applying . parameter estimation techniques. ... • History matching ... Workflow. 8. facies.

Smart Fields, Stanford UniversityIntegrated Operations 200707

Model Order Reduction

4

• PCA, original model size N = 800

full model N = 800n = 1 + 7 = 8n = 2 + 7 = 9n = 8 + 26 = 34n = 23 + 47 = 61

(from Cardoso et al., 2007)

Page 35: Semi-automatic history matching applying parameter ... · Semi-automatic history matching. applying . parameter estimation techniques. ... • History matching ... Workflow. 8. facies.

Smart Fields, Stanford UniversityIntegrated Operations 200707

Model Order Reduction

4

• PCA, original model size N = 800

full model N = 800n = 1 + 7 = 8n = 2 + 7 = 9n = 8 + 26 = 34n = 23 + 47 = 61

(from Cardoso et al., 2007)

Page 36: Semi-automatic history matching applying parameter ... · Semi-automatic history matching. applying . parameter estimation techniques. ... • History matching ... Workflow. 8. facies.

Smart Fields, Stanford UniversityIntegrated Operations 200707

Model Order Reduction

4

• PCA, original model size N = 800

full model N = 800n = 1 + 7 = 8n = 2 + 7 = 9n = 8 + 26 = 34n = 23 + 47 = 61

(from Cardoso et al., 2007)

Page 37: Semi-automatic history matching applying parameter ... · Semi-automatic history matching. applying . parameter estimation techniques. ... • History matching ... Workflow. 8. facies.

Smart Fields, Stanford UniversityIntegrated Operations 200707

Model Order Reduction

4

• PCA, original model size N = 800

full model N = 800n = 1 + 7 = 8n = 2 + 7 = 9n = 8 + 26 = 34n = 23 + 47 = 61

(from Cardoso et al., 2007)

Page 38: Semi-automatic history matching applying parameter ... · Semi-automatic history matching. applying . parameter estimation techniques. ... • History matching ... Workflow. 8. facies.

Smart Fields, Stanford UniversityIntegrated Operations 200707

Outline

O

• Smart Fields Consortium

• The History Matching Problem

• Challenges and Trends

• History Matching at Smart Fields

• Conclusions

Page 39: Semi-automatic history matching applying parameter ... · Semi-automatic history matching. applying . parameter estimation techniques. ... • History matching ... Workflow. 8. facies.

Smart Fields, Stanford UniversityIntegrated Operations 200707

History Matching

5

Page 40: Semi-automatic history matching applying parameter ... · Semi-automatic history matching. applying . parameter estimation techniques. ... • History matching ... Workflow. 8. facies.

Smart Fields, Stanford UniversityIntegrated Operations 200707

History Matching

m**

5

Page 41: Semi-automatic history matching applying parameter ... · Semi-automatic history matching. applying . parameter estimation techniques. ... • History matching ... Workflow. 8. facies.

Smart Fields, Stanford UniversityIntegrated Operations 200707

History Matching

m**observe

5

( )** mO

Page 42: Semi-automatic history matching applying parameter ... · Semi-automatic history matching. applying . parameter estimation techniques. ... • History matching ... Workflow. 8. facies.

Smart Fields, Stanford UniversityIntegrated Operations 200707

m

History Matching

m**observe

( )mO

5

( )** mO

Page 43: Semi-automatic history matching applying parameter ... · Semi-automatic history matching. applying . parameter estimation techniques. ... • History matching ... Workflow. 8. facies.

Smart Fields, Stanford UniversityIntegrated Operations 200707

m

History Matching

( ) ( ) ||mOmO|| minargm̂ **

Mm−=

m**observe ( )** mO

( )mO

5

Page 44: Semi-automatic history matching applying parameter ... · Semi-automatic history matching. applying . parameter estimation techniques. ... • History matching ... Workflow. 8. facies.

Smart Fields, Stanford UniversityIntegrated Operations 200707

A Different Perspective

6

Page 45: Semi-automatic history matching applying parameter ... · Semi-automatic history matching. applying . parameter estimation techniques. ... • History matching ... Workflow. 8. facies.

Smart Fields, Stanford UniversityIntegrated Operations 200707

A Different Perspective

6

• Data filtering– solution quality ~– use observations to improve estimation– when noise/uncertainty is significant

||mm̂|| *−

Page 46: Semi-automatic history matching applying parameter ... · Semi-automatic history matching. applying . parameter estimation techniques. ... • History matching ... Workflow. 8. facies.

Smart Fields, Stanford UniversityIntegrated Operations 200707

A Different Perspective

6

• Data filtering

• Example: Kalman Filter

Page 47: Semi-automatic history matching applying parameter ... · Semi-automatic history matching. applying . parameter estimation techniques. ... • History matching ... Workflow. 8. facies.

Smart Fields, Stanford UniversityIntegrated Operations 200707

A Different Perspective

6

• Data filtering

• Example: Kalman Filter

• Original idea: linear systems only

Page 48: Semi-automatic history matching applying parameter ... · Semi-automatic history matching. applying . parameter estimation techniques. ... • History matching ... Workflow. 8. facies.

Smart Fields, Stanford UniversityIntegrated Operations 200707

A Different Perspective

6

• Data filtering

• Example: Kalman Filter

• Original idea: linear systems only

• Extended Kalman Filter– linearization around current estimate

– problematic with high nonlinearities

Page 49: Semi-automatic history matching applying parameter ... · Semi-automatic history matching. applying . parameter estimation techniques. ... • History matching ... Workflow. 8. facies.

Smart Fields, Stanford UniversityIntegrated Operations 200707

A Different Perspective

6

• Data filtering

• Example: Kalman Filter

• Original idea: linear systems only

• Extended Kalman Filter

• Ensemble Kalman Filter

Page 50: Semi-automatic history matching applying parameter ... · Semi-automatic history matching. applying . parameter estimation techniques. ... • History matching ... Workflow. 8. facies.

Smart Fields, Stanford UniversityIntegrated Operations 200707

A Different Perspective

6

• Data filtering

• Example: Kalman Filter

• Original idea: linear systems only

• Extended Kalman Filter

• Ensemble Kalman Filter– see next presentation!

Page 51: Semi-automatic history matching applying parameter ... · Semi-automatic history matching. applying . parameter estimation techniques. ... • History matching ... Workflow. 8. facies.

Smart Fields, Stanford UniversityIntegrated Operations 200707

Optimization

7

Page 52: Semi-automatic history matching applying parameter ... · Semi-automatic history matching. applying . parameter estimation techniques. ... • History matching ... Workflow. 8. facies.

Smart Fields, Stanford UniversityIntegrated Operations 200707

Optimization

7

• Local optimization

Page 53: Semi-automatic history matching applying parameter ... · Semi-automatic history matching. applying . parameter estimation techniques. ... • History matching ... Workflow. 8. facies.

Smart Fields, Stanford UniversityIntegrated Operations 200707

Optimization

7

• Local optimization– exact / approximate / no gradients

Page 54: Semi-automatic history matching applying parameter ... · Semi-automatic history matching. applying . parameter estimation techniques. ... • History matching ... Workflow. 8. facies.

Smart Fields, Stanford UniversityIntegrated Operations 200707

Optimization

7

• Local optimization– exact / approximate / no gradients– gradients are faster

Page 55: Semi-automatic history matching applying parameter ... · Semi-automatic history matching. applying . parameter estimation techniques. ... • History matching ... Workflow. 8. facies.

Smart Fields, Stanford UniversityIntegrated Operations 200707

Optimization

7

• Local optimization– exact / approximate / no gradients– gradients are faster– no gradients are robust

Page 56: Semi-automatic history matching applying parameter ... · Semi-automatic history matching. applying . parameter estimation techniques. ... • History matching ... Workflow. 8. facies.

Smart Fields, Stanford UniversityIntegrated Operations 200707

Optimization

7

• Local optimization– exact / approximate / no gradients– gradients are faster– no gradients are robust– exact non trivial implementation

Page 57: Semi-automatic history matching applying parameter ... · Semi-automatic history matching. applying . parameter estimation techniques. ... • History matching ... Workflow. 8. facies.

Smart Fields, Stanford UniversityIntegrated Operations 200707

Optimization

7

• Local optimization– exact / approximate / no gradients– gradients are faster– no gradients are robust– exact non trivial implementation– direct methods easier but much slower

Page 58: Semi-automatic history matching applying parameter ... · Semi-automatic history matching. applying . parameter estimation techniques. ... • History matching ... Workflow. 8. facies.

Smart Fields, Stanford UniversityIntegrated Operations 200707

Optimization

7

• Local optimization

• Global optimization

Page 59: Semi-automatic history matching applying parameter ... · Semi-automatic history matching. applying . parameter estimation techniques. ... • History matching ... Workflow. 8. facies.

Smart Fields, Stanford UniversityIntegrated Operations 200707

Optimization

7

• Local optimization

• Global optimization– when gradients make no sense

Page 60: Semi-automatic history matching applying parameter ... · Semi-automatic history matching. applying . parameter estimation techniques. ... • History matching ... Workflow. 8. facies.

Smart Fields, Stanford UniversityIntegrated Operations 200707

Optimization

7

• Local optimization

• Global optimization– when gradients make no sense

(from Onwunalu et al., 2007)

Page 61: Semi-automatic history matching applying parameter ... · Semi-automatic history matching. applying . parameter estimation techniques. ... • History matching ... Workflow. 8. facies.

Smart Fields, Stanford UniversityIntegrated Operations 200707

Optimization

7

• Local optimization

• Global optimization– when gradients make no sense

Page 62: Semi-automatic history matching applying parameter ... · Semi-automatic history matching. applying . parameter estimation techniques. ... • History matching ... Workflow. 8. facies.

Smart Fields, Stanford UniversityIntegrated Operations 200707

Optimization

7

• Local optimization

• Global optimization– when gradients make no sense– most of them stochastic

Page 63: Semi-automatic history matching applying parameter ... · Semi-automatic history matching. applying . parameter estimation techniques. ... • History matching ... Workflow. 8. facies.

Smart Fields, Stanford UniversityIntegrated Operations 200707

Optimization

7

• Local optimization

• Global optimization– when gradients make no sense– most of them stochastic– DIRECT, genetic, particle swarm,

differential evolution, simulated annealing

Page 64: Semi-automatic history matching applying parameter ... · Semi-automatic history matching. applying . parameter estimation techniques. ... • History matching ... Workflow. 8. facies.

Smart Fields, Stanford UniversityIntegrated Operations 200707

Optimization

7

• Local optimization

• Global optimization– when gradients make no sense– most of them stochastic– DIRECT, genetic, particle swarm,

differential evolution, simulated annealing– amenable to parallelization

Page 65: Semi-automatic history matching applying parameter ... · Semi-automatic history matching. applying . parameter estimation techniques. ... • History matching ... Workflow. 8. facies.

Smart Fields, Stanford UniversityIntegrated Operations 200707

Optimization

7

• Local optimization

• Global optimization

• Local faster than global

Page 66: Semi-automatic history matching applying parameter ... · Semi-automatic history matching. applying . parameter estimation techniques. ... • History matching ... Workflow. 8. facies.

Smart Fields, Stanford UniversityIntegrated Operations 200707

Optimization

7

• Local optimization

• Global optimization

• Local faster than global

• Hybridization

Page 67: Semi-automatic history matching applying parameter ... · Semi-automatic history matching. applying . parameter estimation techniques. ... • History matching ... Workflow. 8. facies.

Smart Fields, Stanford UniversityIntegrated Operations 200707

Workflow

8

Page 68: Semi-automatic history matching applying parameter ... · Semi-automatic history matching. applying . parameter estimation techniques. ... • History matching ... Workflow. 8. facies.

Smart Fields, Stanford UniversityIntegrated Operations 200707

m

Workflow

8

facies

Page 69: Semi-automatic history matching applying parameter ... · Semi-automatic history matching. applying . parameter estimation techniques. ... • History matching ... Workflow. 8. facies.

Smart Fields, Stanford UniversityIntegrated Operations 200707

Workflow

8

facies

rock properties

m

Page 70: Semi-automatic history matching applying parameter ... · Semi-automatic history matching. applying . parameter estimation techniques. ... • History matching ... Workflow. 8. facies.

Smart Fields, Stanford UniversityIntegrated Operations 200707

Workflow

8

facies

O1 (m)

O2 (m)

rock properties

m

Page 71: Semi-automatic history matching applying parameter ... · Semi-automatic history matching. applying . parameter estimation techniques. ... • History matching ... Workflow. 8. facies.

Smart Fields, Stanford UniversityIntegrated Operations 200707

Workflow

8

facies

O1 (m)

O2 (m)

m**

rock properties

m

Page 72: Semi-automatic history matching applying parameter ... · Semi-automatic history matching. applying . parameter estimation techniques. ... • History matching ... Workflow. 8. facies.

Smart Fields, Stanford UniversityIntegrated Operations 200707

Workflow

8

facies

O1 (m)

O2 (m)||.||

m**

rock properties

m

Page 73: Semi-automatic history matching applying parameter ... · Semi-automatic history matching. applying . parameter estimation techniques. ... • History matching ... Workflow. 8. facies.

Smart Fields, Stanford UniversityIntegrated Operations 200707

Workflow

8

facies

O1 (m)

O2 (m)||.||

m**

tooptimizerrock

propertiesm

Page 74: Semi-automatic history matching applying parameter ... · Semi-automatic history matching. applying . parameter estimation techniques. ... • History matching ... Workflow. 8. facies.

Smart Fields, Stanford UniversityIntegrated Operations 200707

Workflow

8

facies

flow

seis||.||

m**

tooptimizerrock

propertiesm

Page 75: Semi-automatic history matching applying parameter ... · Semi-automatic history matching. applying . parameter estimation techniques. ... • History matching ... Workflow. 8. facies.

Smart Fields, Stanford UniversityIntegrated Operations 200707

Workflow

8

facies

flow

seis||.||

m**

tooptimizer

manyparameters

rock properties

m

Page 76: Semi-automatic history matching applying parameter ... · Semi-automatic history matching. applying . parameter estimation techniques. ... • History matching ... Workflow. 8. facies.

Smart Fields, Stanford UniversityIntegrated Operations 200707

Workflow

8

facies

flow

seis||.||KPCA

m**

tooptimizer

manyparameters

fewerparameters

rock properties

Page 77: Semi-automatic history matching applying parameter ... · Semi-automatic history matching. applying . parameter estimation techniques. ... • History matching ... Workflow. 8. facies.

Smart Fields, Stanford UniversityIntegrated Operations 200707

Workflow

8

facies

flow

seis||.||KPCA

m**

tooptimizer

manyparameters

lessparameters

rock properties

gradients

Page 78: Semi-automatic history matching applying parameter ... · Semi-automatic history matching. applying . parameter estimation techniques. ... • History matching ... Workflow. 8. facies.

Smart Fields, Stanford UniversityIntegrated Operations 200707

Outline

O

• Smart Fields Consortium

• The History Matching Problem

• Challenges and Trends

• History Matching at Smart Fields

• Conclusions

Page 79: Semi-automatic history matching applying parameter ... · Semi-automatic history matching. applying . parameter estimation techniques. ... • History matching ... Workflow. 8. facies.

Smart Fields, Stanford UniversityIntegrated Operations 200707

Challenges and Trends

9

Page 80: Semi-automatic history matching applying parameter ... · Semi-automatic history matching. applying . parameter estimation techniques. ... • History matching ... Workflow. 8. facies.

Smart Fields, Stanford UniversityIntegrated Operations 200707

Challenges and Trends

9

• Non-uniqueness in optimization

Page 81: Semi-automatic history matching applying parameter ... · Semi-automatic history matching. applying . parameter estimation techniques. ... • History matching ... Workflow. 8. facies.

Smart Fields, Stanford UniversityIntegrated Operations 200707

Challenges and Trends

9

• Non-uniqueness in optimization

0.0

0.2

0.4

0.6

0.8

1.0

Wat

erC

ut

0 200 400 600 800 1000tim e, days

a a a a a a a a a

a

a

a

a

a

a

a

a

a

a

a

a

aa

a

a

a

aa

aa a a a a a a a a a a a a a

a a a a a a a a a

a

a

a

a

a

a

a

a

a

a

a

a

a

aa

aa

aa

aa

aa

aa a a a a a a a a a

a

a

In itia l Perm eab

Wat

er c

ut

t (days)

(from Caers et al., 2002)

Page 82: Semi-automatic history matching applying parameter ... · Semi-automatic history matching. applying . parameter estimation techniques. ... • History matching ... Workflow. 8. facies.

Smart Fields, Stanford UniversityIntegrated Operations 200707

Challenges and Trends

9

• Non-uniqueness in optimization

0.0

0.2

0.4

0.6

0.8

1.0

Wat

erC

ut

0 200 400 600 800 1000tim e, days

a a a a a a a a a

a

a

a

a

a

a

a

a

a

a

a

a

aa

a

a

a

aa

aa a a a a a a a a a a a a a

a a a a a a a a a

a

a

a

a

a

a

a

a

a

a

a

a

a

aa

aa

aa

aa

aa

aa a a a a a a a a a

a

a

In itia l Perm eab

Wat

er c

ut

t (days)

East

Nor

th

0.0 50.00.0

50.0

0.0

200.0

400.0

600.0

800.0

1000.0

East

Nor

th

0.0 50.00.0

50.0

0.0

200.0

400.0

600.0

800.0

1000.0

East

Nor

th

0.0 50.00.0

50.0

0.0

200.0

400.0

600.0

800.0

1000.0

East

Nor

th

0.0 50.00.0

50.0

0.0

200.0

400.0

600.0

800.0

1000.0

(from Caers et al., 2002)

Page 83: Semi-automatic history matching applying parameter ... · Semi-automatic history matching. applying . parameter estimation techniques. ... • History matching ... Workflow. 8. facies.

Smart Fields, Stanford UniversityIntegrated Operations 200707

Challenges and Trends

9

• Non-uniqueness in optimization

0.0

0.2

0.4

0.6

0.8

1.0

Wat

erC

ut

0 200 400 600 800 1000tim e, days

a a a a a a a a a

a

a

a

a

a

a

a

a

a

a

a

a

aa

a

a

a

aa

aa a a a a a a a a a a a a a

a a a a a a a a a

a

a

a

a

a

a

a

a

a

a

a

a

a

aa

aa

aa

aa

aa

aa a a a a a a a a a

a

a

In itia l Perm eab

Wat

er c

ut

t (days)

East

Nor

th

0.0 50.00.0

50.0

0.0

200.0

400.0

600.0

800.0

1000.0

East

Nor

th

0.0 50.00.0

50.0

0.0

200.0

400.0

600.0

800.0

1000.0

East

Nor

th

0.0 50.00.0

50.0

0.0

200.0

400.0

600.0

800.0

1000.0

East

Nor

th

0.0 50.00.0

50.0

0.0

200.0

400.0

600.0

800.0

1000.0

(from Caers et al., 2002)

Page 84: Semi-automatic history matching applying parameter ... · Semi-automatic history matching. applying . parameter estimation techniques. ... • History matching ... Workflow. 8. facies.

Smart Fields, Stanford UniversityIntegrated Operations 200707

Challenges and Trends

9

• Non-uniqueness in optimization– select a solution that honors geology

– KPCA

Page 85: Semi-automatic history matching applying parameter ... · Semi-automatic history matching. applying . parameter estimation techniques. ... • History matching ... Workflow. 8. facies.

Smart Fields, Stanford UniversityIntegrated Operations 200707

Challenges and Trends

9

• Non-uniqueness in optimization– honors geology, KPCA

• Large-scale optimization– from thousands to millions of variables

– reduce number of variables

– KPCA

Page 86: Semi-automatic history matching applying parameter ... · Semi-automatic history matching. applying . parameter estimation techniques. ... • History matching ... Workflow. 8. facies.

Smart Fields, Stanford UniversityIntegrated Operations 200707

Challenges and Trends

9

• Non-uniqueness in optimization– honors geology, KPCA

• Large-scale optimization– reduce number of variables, KPCA

Page 87: Semi-automatic history matching applying parameter ... · Semi-automatic history matching. applying . parameter estimation techniques. ... • History matching ... Workflow. 8. facies.

Smart Fields, Stanford UniversityIntegrated Operations 200707

Challenges and Trends

9

• Non-uniqueness in optimization– honors geology, KPCA

• Large-scale optimization– reduce number of variables, KPCA

• Cost functions expensive to compute– reduce number of function calls, gradients

Page 88: Semi-automatic history matching applying parameter ... · Semi-automatic history matching. applying . parameter estimation techniques. ... • History matching ... Workflow. 8. facies.

Smart Fields, Stanford UniversityIntegrated Operations 200707

More Trends

10

Page 89: Semi-automatic history matching applying parameter ... · Semi-automatic history matching. applying . parameter estimation techniques. ... • History matching ... Workflow. 8. facies.

Smart Fields, Stanford UniversityIntegrated Operations 200707

More Trends

10

• Global methods

Page 90: Semi-automatic history matching applying parameter ... · Semi-automatic history matching. applying . parameter estimation techniques. ... • History matching ... Workflow. 8. facies.

Smart Fields, Stanford UniversityIntegrated Operations 200707

More Trends

10

• Global methods– dealing with uncertainty and noise

Page 91: Semi-automatic history matching applying parameter ... · Semi-automatic history matching. applying . parameter estimation techniques. ... • History matching ... Workflow. 8. facies.

Smart Fields, Stanford UniversityIntegrated Operations 200707

More Trends

10

• Global methods– dealing with uncertainty and noise

– when a good initial guess is not available

Page 92: Semi-automatic history matching applying parameter ... · Semi-automatic history matching. applying . parameter estimation techniques. ... • History matching ... Workflow. 8. facies.

Smart Fields, Stanford UniversityIntegrated Operations 200707

More Trends

10

• Global methods– dealing with uncertainty and noise

– when a good initial guess is not available

– combined with proxies

Page 93: Semi-automatic history matching applying parameter ... · Semi-automatic history matching. applying . parameter estimation techniques. ... • History matching ... Workflow. 8. facies.

Smart Fields, Stanford UniversityIntegrated Operations 200707

More Trends

10

• Global methods– dealing with uncertainty and noise

– when a good initial guess is not available

– combined with proxies

– hybridize with local methods

Page 94: Semi-automatic history matching applying parameter ... · Semi-automatic history matching. applying . parameter estimation techniques. ... • History matching ... Workflow. 8. facies.

Smart Fields, Stanford UniversityIntegrated Operations 200707

More Trends

10

• Global methods– dealing with uncertainty and noise

– when a good initial guess is not available

– combined with proxies

– hybridize with local methods

• Data filtering

Page 95: Semi-automatic history matching applying parameter ... · Semi-automatic history matching. applying . parameter estimation techniques. ... • History matching ... Workflow. 8. facies.

Smart Fields, Stanford UniversityIntegrated Operations 200707

Workflow

8

facies

flow

seis||.||KPCA

m**

tooptimizer

manyparameters

lessparameters

rock properties

gradients

Page 96: Semi-automatic history matching applying parameter ... · Semi-automatic history matching. applying . parameter estimation techniques. ... • History matching ... Workflow. 8. facies.

Smart Fields, Stanford UniversityIntegrated Operations 200707

Kernel PCA

12

Page 97: Semi-automatic history matching applying parameter ... · Semi-automatic history matching. applying . parameter estimation techniques. ... • History matching ... Workflow. 8. facies.

Smart Fields, Stanford UniversityIntegrated Operations 200707

Kernel PCA

11

• PCA captures linear dependence

Page 98: Semi-automatic history matching applying parameter ... · Semi-automatic history matching. applying . parameter estimation techniques. ... • History matching ... Workflow. 8. facies.

Smart Fields, Stanford UniversityIntegrated Operations 200707

Kernel PCA

11

• PCA captures linear dependence

Page 99: Semi-automatic history matching applying parameter ... · Semi-automatic history matching. applying . parameter estimation techniques. ... • History matching ... Workflow. 8. facies.

Smart Fields, Stanford UniversityIntegrated Operations 200707

Kernel PCA

11

• PCA captures linear dependence

Page 100: Semi-automatic history matching applying parameter ... · Semi-automatic history matching. applying . parameter estimation techniques. ... • History matching ... Workflow. 8. facies.

Smart Fields, Stanford UniversityIntegrated Operations 200707

Kernel PCA

11

• PCA captures linear dependence

Page 101: Semi-automatic history matching applying parameter ... · Semi-automatic history matching. applying . parameter estimation techniques. ... • History matching ... Workflow. 8. facies.

Smart Fields, Stanford UniversityIntegrated Operations 200707

Kernel PCA

11

• PCA captures linear dependence

Page 102: Semi-automatic history matching applying parameter ... · Semi-automatic history matching. applying . parameter estimation techniques. ... • History matching ... Workflow. 8. facies.

Smart Fields, Stanford UniversityIntegrated Operations 200707

Kernel PCA

11

• PCA captures linear dependence

⎪⎭

⎪⎬⎫

⎪⎩

⎪⎨⎧

+=

=

21

212

1

12

yxyxyarctanx

Page 103: Semi-automatic history matching applying parameter ... · Semi-automatic history matching. applying . parameter estimation techniques. ... • History matching ... Workflow. 8. facies.

Smart Fields, Stanford UniversityIntegrated Operations 200707

Kernel PCA

11

• PCA captures linear dependence

• Nonlinear is more flexible

Page 104: Semi-automatic history matching applying parameter ... · Semi-automatic history matching. applying . parameter estimation techniques. ... • History matching ... Workflow. 8. facies.

Smart Fields, Stanford UniversityIntegrated Operations 200707

Kernel PCA

11

• PCA captures linear dependence

• Nonlinear is more flexible

• PCA preserves 2nd order statistics

Page 105: Semi-automatic history matching applying parameter ... · Semi-automatic history matching. applying . parameter estimation techniques. ... • History matching ... Workflow. 8. facies.

Smart Fields, Stanford UniversityIntegrated Operations 200707

Kernel PCA

11

• PCA captures linear dependence

• Nonlinear is more flexible

• PCA preserves 2nd order statistics

requires more than 2nd order

Page 106: Semi-automatic history matching applying parameter ... · Semi-automatic history matching. applying . parameter estimation techniques. ... • History matching ... Workflow. 8. facies.

Smart Fields, Stanford UniversityIntegrated Operations 200707

Kernel PCA

11

• PCA captures linear dependence

• Nonlinear is more flexible

• PCA preserves 2nd order statistics

Page 107: Semi-automatic history matching applying parameter ... · Semi-automatic history matching. applying . parameter estimation techniques. ... • History matching ... Workflow. 8. facies.

Smart Fields, Stanford UniversityIntegrated Operations 200707

Kernel PCA

11

• PCA captures linear dependence

• Nonlinear is more flexible

• PCA preserves 2nd order statistics

• Higher dimensions preserve

higher order statistics

Page 108: Semi-automatic history matching applying parameter ... · Semi-automatic history matching. applying . parameter estimation techniques. ... • History matching ... Workflow. 8. facies.

Smart Fields, Stanford UniversityIntegrated Operations 200707

KPCA in Action(from Sarma et al., 2006)

12

Page 109: Semi-automatic history matching applying parameter ... · Semi-automatic history matching. applying . parameter estimation techniques. ... • History matching ... Workflow. 8. facies.

Smart Fields, Stanford UniversityIntegrated Operations 200707

KPCA in Action

12

-1 0 1 20

0.5

1Approximate pdf

-2 -1 0 1 2 30

0.2

0.4

0.6

0.8Approximate pdf

-0.5

0

0.5

1

1.5 Perm Field from KLE

10 20 30 40

10

20

30

40

0

1

2Perm Field from KPCA

10 20 30 40

10

20

30

40

PCA analysis (n = 30)

(from Sarma et al., 2006)

KPCA analysis (n = 30)

Page 110: Semi-automatic history matching applying parameter ... · Semi-automatic history matching. applying . parameter estimation techniques. ... • History matching ... Workflow. 8. facies.

Smart Fields, Stanford UniversityIntegrated Operations 200707

KPCA in Action

12

-1 0 1 20

0.5

1Approximate pdf

-2 -1 0 1 2 30

0.2

0.4

0.6

0.8Approximate pdf

-0.5

0

0.5

1

1.5 Perm Field from KLE

10 20 30 40

10

20

30

40

0

1

2Perm Field from KPCA

10 20 30 40

10

20

30

40

PCA analysis (n = 30)

(from Sarma et al., 2006)

KPCA analysis (n = 30)

Page 111: Semi-automatic history matching applying parameter ... · Semi-automatic history matching. applying . parameter estimation techniques. ... • History matching ... Workflow. 8. facies.

Smart Fields, Stanford UniversityIntegrated Operations 200707

KPCA in Action

12

(from Sarma et al., 2006)

Page 112: Semi-automatic history matching applying parameter ... · Semi-automatic history matching. applying . parameter estimation techniques. ... • History matching ... Workflow. 8. facies.

Smart Fields, Stanford UniversityIntegrated Operations 200707

-1 0 1 20

0.5

1Approximate pdf

-2 -1 0 1 2 30

0.2

0.4

0.6

0.8Approximate pdf

-0.5

0

0.5

1

Perm Field from KLE

10 20 30 40

10

20

30

40

-0.5

0

0.5

1

1.5

Perm Field from KPCA

10 20 30 40

10

20

30

40

KPCA in Action

12

PCA analysis (n = 30)

(from Sarma et al., 2006)

KPCA analysis (n = 30)

Page 113: Semi-automatic history matching applying parameter ... · Semi-automatic history matching. applying . parameter estimation techniques. ... • History matching ... Workflow. 8. facies.

Smart Fields, Stanford UniversityIntegrated Operations 200707

-1 0 1 20

0.5

1Approximate pdf

-2 -1 0 1 2 30

0.2

0.4

0.6

0.8Approximate pdf

-0.5

0

0.5

1

Perm Field from KLE

10 20 30 40

10

20

30

40

-0.5

0

0.5

1

1.5

Perm Field from KPCA

10 20 30 40

10

20

30

40

KPCA in Action

12

PCA analysis (n = 30)

(from Sarma et al., 2006)

KPCA analysis (n = 30)

Page 114: Semi-automatic history matching applying parameter ... · Semi-automatic history matching. applying . parameter estimation techniques. ... • History matching ... Workflow. 8. facies.

Smart Fields, Stanford UniversityIntegrated Operations 200707

Flow Gradients

13

Page 115: Semi-automatic history matching applying parameter ... · Semi-automatic history matching. applying . parameter estimation techniques. ... • History matching ... Workflow. 8. facies.

Smart Fields, Stanford UniversityIntegrated Operations 200707

Flow Gradients

13

• Cost function = flow + seismic

Page 116: Semi-automatic history matching applying parameter ... · Semi-automatic history matching. applying . parameter estimation techniques. ... • History matching ... Workflow. 8. facies.

Smart Fields, Stanford UniversityIntegrated Operations 200707

Flow Gradients

13

• Cost function = flow + seismic

• Flow cost function

( ) ( )∑−

=

=1N

0nn mLmJ m: model

( ) ( ) ||mOmO|| J(m) **11 −=

Page 117: Semi-automatic history matching applying parameter ... · Semi-automatic history matching. applying . parameter estimation techniques. ... • History matching ... Workflow. 8. facies.

Smart Fields, Stanford UniversityIntegrated Operations 200707

Flow Gradients

13

• Cost function = flow + seismic

• Flow cost function

• Constraints (reservoir flow equations)

( ) 0m,x,xg n1nn =+ x: states

( ) ( )∑−

=

=1N

0nn mLmJ m: model

Page 118: Semi-automatic history matching applying parameter ... · Semi-automatic history matching. applying . parameter estimation techniques. ... • History matching ... Workflow. 8. facies.

Smart Fields, Stanford UniversityIntegrated Operations 200707

Flow Gradients

13

• Cost function = flow + seismic

• Cost function and constraints

• Optimality

( ) 0m,x,xg ,0mJ

n1nnA ==

∂∂

+

( ) ( ) ( )m,x,xg m,xL,mJ n1nnT

1n

1N

0n1nnA ++

=+ λ+=λ ∑

( ) ( )λ=λλ

,mJ minargˆ,m̂ A,m

ff

Page 119: Semi-automatic history matching applying parameter ... · Semi-automatic history matching. applying . parameter estimation techniques. ... • History matching ... Workflow. 8. facies.

Smart Fields, Stanford UniversityIntegrated Operations 200707

Flow Gradients

13

• Cost function = flow + seismic

• Cost function and constraints

• Optimality

( ) 0m,x,xg ,0mJ

n1nnA ==

∂∂

+

( ) ( ) ( )m,x,xg m,xL,mJ n1nnT

1n

1N

0n1nnA ++

=+ λ+=λ ∑

( ) ( )λ=λλ

,mJ minargˆ,m̂ A,m

ff

Page 120: Semi-automatic history matching applying parameter ... · Semi-automatic history matching. applying . parameter estimation techniques. ... • History matching ... Workflow. 8. facies.

Smart Fields, Stanford UniversityIntegrated Operations 200707

Adjoint Equations

• Adjoint equations directly obtained from reservoir simulator

• Store Jacobians: and

• Nontrivial implementation

• Other nonlinear constraints can be considered

n

1n

xg∂∂ −

n

n

xg∂∂

14

Page 121: Semi-automatic history matching applying parameter ... · Semi-automatic history matching. applying . parameter estimation techniques. ... • History matching ... Workflow. 8. facies.

Smart Fields, Stanford UniversityIntegrated Operations 200707

Outline

O

• Smart Fields Consortium

• The History Matching Problem

• Challenges and Trends

• History Matching at Smart Fields

• Conclusions

Page 122: Semi-automatic history matching applying parameter ... · Semi-automatic history matching. applying . parameter estimation techniques. ... • History matching ... Workflow. 8. facies.

Smart Fields, Stanford UniversityIntegrated Operations 200707

Probability Perturbation(from Caers et al., 2003)

15

Page 123: Semi-automatic history matching applying parameter ... · Semi-automatic history matching. applying . parameter estimation techniques. ... • History matching ... Workflow. 8. facies.

Smart Fields, Stanford UniversityIntegrated Operations 200707

Probability Perturbation

• Global Optimization

(from Caers et al., 2003)

15

Page 124: Semi-automatic history matching applying parameter ... · Semi-automatic history matching. applying . parameter estimation techniques. ... • History matching ... Workflow. 8. facies.

Smart Fields, Stanford UniversityIntegrated Operations 200707

Probability Perturbation

• Global Optimization

high

low

cost function

(from Caers et al., 2003)

15

Page 125: Semi-automatic history matching applying parameter ... · Semi-automatic history matching. applying . parameter estimation techniques. ... • History matching ... Workflow. 8. facies.

Smart Fields, Stanford UniversityIntegrated Operations 200707

Probability Perturbation

• Global Optimization

high

low

cost function

(from Caers et al., 2003)

15

Page 126: Semi-automatic history matching applying parameter ... · Semi-automatic history matching. applying . parameter estimation techniques. ... • History matching ... Workflow. 8. facies.

Smart Fields, Stanford UniversityIntegrated Operations 200707

Probability Perturbation

• Global Optimization

high

low

cost function

random

(from Caers et al., 2003)

15

Page 127: Semi-automatic history matching applying parameter ... · Semi-automatic history matching. applying . parameter estimation techniques. ... • History matching ... Workflow. 8. facies.

Smart Fields, Stanford UniversityIntegrated Operations 200707

Probability Perturbation

• Global Optimization

high

low

cost function

random

(from Caers et al., 2003)

15

Page 128: Semi-automatic history matching applying parameter ... · Semi-automatic history matching. applying . parameter estimation techniques. ... • History matching ... Workflow. 8. facies.

Smart Fields, Stanford UniversityIntegrated Operations 200707

Probability Perturbation

• Global Optimization

high

low

cost function

random

(from Caers et al., 2003)

15

Page 129: Semi-automatic history matching applying parameter ... · Semi-automatic history matching. applying . parameter estimation techniques. ... • History matching ... Workflow. 8. facies.

Smart Fields, Stanford UniversityIntegrated Operations 200707

Probability Perturbation

• Global Optimization

high

low

cost function

curve optimumrandom

(from Caers et al., 2003)

15

Page 130: Semi-automatic history matching applying parameter ... · Semi-automatic history matching. applying . parameter estimation techniques. ... • History matching ... Workflow. 8. facies.

Smart Fields, Stanford UniversityIntegrated Operations 200707

Probability Perturbation

• Global Optimization

high

low

cost function

curve optimumrandom

(from Caers et al., 2003)

15

Page 131: Semi-automatic history matching applying parameter ... · Semi-automatic history matching. applying . parameter estimation techniques. ... • History matching ... Workflow. 8. facies.

Smart Fields, Stanford UniversityIntegrated Operations 200707

Probability Perturbation

• Global Optimization

high

low

cost function

curve optimumrandom

(from Caers et al., 2003)

15

Page 132: Semi-automatic history matching applying parameter ... · Semi-automatic history matching. applying . parameter estimation techniques. ... • History matching ... Workflow. 8. facies.

Smart Fields, Stanford UniversityIntegrated Operations 200707

Probability Perturbation

• Global Optimization

high

low

cost function

curve optimumrandom

(from Caers et al., 2003)

15

Page 133: Semi-automatic history matching applying parameter ... · Semi-automatic history matching. applying . parameter estimation techniques. ... • History matching ... Workflow. 8. facies.

Smart Fields, Stanford UniversityIntegrated Operations 200707

Probability Perturbation

• Global Optimization

high

low

cost function

curve optimumrandom

(from Caers et al., 2003)

15

Page 134: Semi-automatic history matching applying parameter ... · Semi-automatic history matching. applying . parameter estimation techniques. ... • History matching ... Workflow. 8. facies.

Smart Fields, Stanford UniversityIntegrated Operations 200707

Probability Perturbation

• Global Optimization

high

low

cost function

curve optimumrandom

(from Caers et al., 2003)

15

Page 135: Semi-automatic history matching applying parameter ... · Semi-automatic history matching. applying . parameter estimation techniques. ... • History matching ... Workflow. 8. facies.

Smart Fields, Stanford UniversityIntegrated Operations 200707

Probability Perturbation

• Global Optimization

high

low

cost function

curve optimumrandom

(from Caers et al., 2003)

15

Page 136: Semi-automatic history matching applying parameter ... · Semi-automatic history matching. applying . parameter estimation techniques. ... • History matching ... Workflow. 8. facies.

Smart Fields, Stanford UniversityIntegrated Operations 200707

Probability Perturbation

• Global Optimization

high

low

cost function

curve optimumrandom

(from Caers et al., 2003)

15

Page 137: Semi-automatic history matching applying parameter ... · Semi-automatic history matching. applying . parameter estimation techniques. ... • History matching ... Workflow. 8. facies.

Smart Fields, Stanford UniversityIntegrated Operations 200707

Probability Perturbation

• Global Optimization

high

low

cost function

curve optimumrandom

(from Caers et al., 2003)

15

Page 138: Semi-automatic history matching applying parameter ... · Semi-automatic history matching. applying . parameter estimation techniques. ... • History matching ... Workflow. 8. facies.

Smart Fields, Stanford UniversityIntegrated Operations 200707

Probability Perturbation

• Global Optimization

high

low

cost function

curve optimumrandom

(from Caers et al., 2003)

15

Page 139: Semi-automatic history matching applying parameter ... · Semi-automatic history matching. applying . parameter estimation techniques. ... • History matching ... Workflow. 8. facies.

Smart Fields, Stanford UniversityIntegrated Operations 200707

Probability Perturbation

• Global Optimization

high

low

cost function

curve optimumrandom

(from Caers et al., 2003)

15

Page 140: Semi-automatic history matching applying parameter ... · Semi-automatic history matching. applying . parameter estimation techniques. ... • History matching ... Workflow. 8. facies.

Smart Fields, Stanford UniversityIntegrated Operations 200707

Probability Perturbation

• Global Optimization

(from Caers et al., 2003)

16

Page 141: Semi-automatic history matching applying parameter ... · Semi-automatic history matching. applying . parameter estimation techniques. ... • History matching ... Workflow. 8. facies.

Smart Fields, Stanford UniversityIntegrated Operations 200707

Probability Perturbation

• Global Optimization

• Related projects (Jef Caers)– History matching based on multiple alternative

geological scenarios

– Modeling of hydrogeological deposits constrained to pressure data

– Integration of 4D seismics and production data

– just to name a few...

(from Caers et al., 2003)

16

Page 142: Semi-automatic history matching applying parameter ... · Semi-automatic history matching. applying . parameter estimation techniques. ... • History matching ... Workflow. 8. facies.

Smart Fields, Stanford UniversityIntegrated Operations 200707

Closed-Loop Management

17

(from Sarma et al., 2006)

Page 143: Semi-automatic history matching applying parameter ... · Semi-automatic history matching. applying . parameter estimation techniques. ... • History matching ... Workflow. 8. facies.

Smart Fields, Stanford UniversityIntegrated Operations 200707

Closed-Loop Management

• Model (sector)

– 32x46x8 (~12000) cells– 3 injectors and 4 producers, BHP control

(from Sarma et al., 2006)

17

Page 144: Semi-automatic history matching applying parameter ... · Semi-automatic history matching. applying . parameter estimation techniques. ... • History matching ... Workflow. 8. facies.

Smart Fields, Stanford UniversityIntegrated Operations 200707

Closed-Loop Management

• Control problem– permeability field unknown (model update)– maximize NPV in ~ 8 years (costly water)

(from Sarma et al., 2006)

17

Page 145: Semi-automatic history matching applying parameter ... · Semi-automatic history matching. applying . parameter estimation techniques. ... • History matching ... Workflow. 8. facies.

Smart Fields, Stanford UniversityIntegrated Operations 200707

Closed-Loop Management

• Constraints– bounds for controls– total injection ≤

20,000 STBD, watercut ≤

0.95

(from Sarma et al., 2006)

17

Page 146: Semi-automatic history matching applying parameter ... · Semi-automatic history matching. applying . parameter estimation techniques. ... • History matching ... Workflow. 8. facies.

Smart Fields, Stanford UniversityIntegrated Operations 200707

Closed-Loop Management

• Reference model– producer BHP = 4500 psi– injection water distributed by kh

(from Sarma et al., 2006)

17

Page 147: Semi-automatic history matching applying parameter ... · Semi-automatic history matching. applying . parameter estimation techniques. ... • History matching ... Workflow. 8. facies.

Smart Fields, Stanford UniversityIntegrated Operations 200707

Oil Saturations(from Sarma et al., 2006)

18

Page 148: Semi-automatic history matching applying parameter ... · Semi-automatic history matching. applying . parameter estimation techniques. ... • History matching ... Workflow. 8. facies.

Smart Fields, Stanford UniversityIntegrated Operations 200707

Oil Saturations(from Sarma et al., 2006)

reference optimized(known permeability)

18

Page 149: Semi-automatic history matching applying parameter ... · Semi-automatic history matching. applying . parameter estimation techniques. ... • History matching ... Workflow. 8. facies.

Smart Fields, Stanford UniversityIntegrated Operations 200707

Oil Saturations(from Sarma et al., 2006)

optimized(known permeability)

optimized(unknown permeability)

18

Page 150: Semi-automatic history matching applying parameter ... · Semi-automatic history matching. applying . parameter estimation techniques. ... • History matching ... Workflow. 8. facies.

Smart Fields, Stanford UniversityIntegrated Operations 200707

Production Data(from Sarma et al., 2006)

19

Page 151: Semi-automatic history matching applying parameter ... · Semi-automatic history matching. applying . parameter estimation techniques. ... • History matching ... Workflow. 8. facies.

Smart Fields, Stanford UniversityIntegrated Operations 200707

Production Data(from Sarma et al., 2006)

base case

Cum

ulat

ives

(STB

)

optimized

reference

19

Page 152: Semi-automatic history matching applying parameter ... · Semi-automatic history matching. applying . parameter estimation techniques. ... • History matching ... Workflow. 8. facies.

Smart Fields, Stanford UniversityIntegrated Operations 200707

Production Data(from Sarma et al., 2006)

base case

Cum

ulat

ives

(STB

)

16% increase in oil production

optimized

reference

19

Page 153: Semi-automatic history matching applying parameter ... · Semi-automatic history matching. applying . parameter estimation techniques. ... • History matching ... Workflow. 8. facies.

Smart Fields, Stanford UniversityIntegrated Operations 200707

Production Data(from Sarma et al., 2006)

base case

Cum

ulat

ives

(STB

)

50% decrease in water production

optimized

reference

19

Page 154: Semi-automatic history matching applying parameter ... · Semi-automatic history matching. applying . parameter estimation techniques. ... • History matching ... Workflow. 8. facies.

Smart Fields, Stanford UniversityIntegrated Operations 200707

Production Data(from Sarma et al., 2006)

base case

Cum

ulat

ives

(STB

)

25% increase in NPV

optimized

reference

19

Page 155: Semi-automatic history matching applying parameter ... · Semi-automatic history matching. applying . parameter estimation techniques. ... • History matching ... Workflow. 8. facies.

Smart Fields, Stanford UniversityIntegrated Operations 200707

Integrating Data

20

(joint work with Mukerji and Santos)

Page 156: Semi-automatic history matching applying parameter ... · Semi-automatic history matching. applying . parameter estimation techniques. ... • History matching ... Workflow. 8. facies.

Smart Fields, Stanford UniversityIntegrated Operations 200707

Integrating Data

• Production data and seismics

20

(joint work with Mukerji and Santos)

Page 157: Semi-automatic history matching applying parameter ... · Semi-automatic history matching. applying . parameter estimation techniques. ... • History matching ... Workflow. 8. facies.

Smart Fields, Stanford UniversityIntegrated Operations 200707

Integrating Data

• Production data and seismics

• Stanford VI reservoir

20

(joint work with Mukerji and Santos)

Page 158: Semi-automatic history matching applying parameter ... · Semi-automatic history matching. applying . parameter estimation techniques. ... • History matching ... Workflow. 8. facies.

Smart Fields, Stanford UniversityIntegrated Operations 200707

Integrating Data

• Production data and seismics

• Stanford VI reservoir– three zones

zone 3

20

(joint work with Mukerji and Santos)

Page 159: Semi-automatic history matching applying parameter ... · Semi-automatic history matching. applying . parameter estimation techniques. ... • History matching ... Workflow. 8. facies.

Smart Fields, Stanford UniversityIntegrated Operations 200707

Integrating Data

• Production data and seismics

• Stanford VI reservoir– three zones

– prograding fluvial channel

zone 3

20

(joint work with Mukerji and Santos)

Page 160: Semi-automatic history matching applying parameter ... · Semi-automatic history matching. applying . parameter estimation techniques. ... • History matching ... Workflow. 8. facies.

Smart Fields, Stanford UniversityIntegrated Operations 200707

Integrating Data

• Production data and seismics

• Stanford VI reservoir– three zones

– prograding fluvial channel

– an asymmetric anticline

zone 3

20

(joint work with Mukerji and Santos)

Page 161: Semi-automatic history matching applying parameter ... · Semi-automatic history matching. applying . parameter estimation techniques. ... • History matching ... Workflow. 8. facies.

Smart Fields, Stanford UniversityIntegrated Operations 200707

Integrating Data

• Production data and seismics

• Stanford VI reservoir– three zones

– prograding fluvial channel

– an asymmetric anticline

– 6 million cellszone 3

20

(joint work with Mukerji and Santos)

Page 162: Semi-automatic history matching applying parameter ... · Semi-automatic history matching. applying . parameter estimation techniques. ... • History matching ... Workflow. 8. facies.

Smart Fields, Stanford UniversityIntegrated Operations 200707

Integrating Data

• Production data and seismics

• Stanford VI reservoir– three zones

– prograding fluvial channel

– an asymmetric anticline

– 6 million cells

– 4D seismic data set zone 3

20

(joint work with Mukerji and Santos)

Page 163: Semi-automatic history matching applying parameter ... · Semi-automatic history matching. applying . parameter estimation techniques. ... • History matching ... Workflow. 8. facies.

Smart Fields, Stanford UniversityIntegrated Operations 200707

Integrating Data

• Production data and seismics

• Stanford VI reservoir

20

(joint work with Mukerji and Santos)

Page 164: Semi-automatic history matching applying parameter ... · Semi-automatic history matching. applying . parameter estimation techniques. ... • History matching ... Workflow. 8. facies.

Smart Fields, Stanford UniversityIntegrated Operations 200707

• Production data and seismics

• Stanford VI reservoir

• Tomographies, 4D

Integrating Data

20

(joint work with Mukerji and Santos)

Page 165: Semi-automatic history matching applying parameter ... · Semi-automatic history matching. applying . parameter estimation techniques. ... • History matching ... Workflow. 8. facies.

Smart Fields, Stanford UniversityIntegrated Operations 200707

• Production data and seismics

• Stanford VI reservoir

• Tomographies, 4D

• Flexible optimization– alternating matching production/seismics

– cumulative match from tn to tn+1

Integrating Data

20

(joint work with Mukerji and Santos)

Page 166: Semi-automatic history matching applying parameter ... · Semi-automatic history matching. applying . parameter estimation techniques. ... • History matching ... Workflow. 8. facies.

Smart Fields, Stanford UniversityIntegrated Operations 200707

Preliminary Results

21

(joint work with Mukerji and Santos)

Page 167: Semi-automatic history matching applying parameter ... · Semi-automatic history matching. applying . parameter estimation techniques. ... • History matching ... Workflow. 8. facies.

Smart Fields, Stanford UniversityIntegrated Operations 200707

Preliminary Results

• 20x20x10 sector from zone 3

21

(joint work with Mukerji and Santos)

Page 168: Semi-automatic history matching applying parameter ... · Semi-automatic history matching. applying . parameter estimation techniques. ... • History matching ... Workflow. 8. facies.

Smart Fields, Stanford UniversityIntegrated Operations 200707

Preliminary Results

• 20x20x10 sector from zone 3

• m: facies

21

(joint work with Mukerji and Santos)

Page 169: Semi-automatic history matching applying parameter ... · Semi-automatic history matching. applying . parameter estimation techniques. ... • History matching ... Workflow. 8. facies.

Smart Fields, Stanford UniversityIntegrated Operations 200707

Preliminary Results

• 20x20x10 sector from zone 3

• m: facies

• k, φ: regression from well location– k from Kozeny-Carman’s

21

(joint work with Mukerji and Santos)

Page 170: Semi-automatic history matching applying parameter ... · Semi-automatic history matching. applying . parameter estimation techniques. ... • History matching ... Workflow. 8. facies.

Smart Fields, Stanford UniversityIntegrated Operations 200707

Preliminary Results

• 20x20x10 sector from zone 3

• m: facies

• k, φ: regression from well location

• VP and VS as seismics

21

(joint work with Mukerji and Santos)

Page 171: Semi-automatic history matching applying parameter ... · Semi-automatic history matching. applying . parameter estimation techniques. ... • History matching ... Workflow. 8. facies.

Smart Fields, Stanford UniversityIntegrated Operations 200707

Preliminary Results

• 20x20x10 sector from zone 3

• m: facies

• k, φ: regression from well location

• VP and VS as seismics– Gassmann fluid substitution

– Batzle-Wang fluid properties

21

(joint work with Mukerji and Santos)

Page 172: Semi-automatic history matching applying parameter ... · Semi-automatic history matching. applying . parameter estimation techniques. ... • History matching ... Workflow. 8. facies.

Smart Fields, Stanford UniversityIntegrated Operations 200707

Preliminary Results

• 20x20x10 sector from zone 3

• m: facies

• k, φ: regression from well location

• VP and VS as seismics

• PCA: 30 coefficients retained

21

(joint work with Mukerji and Santos)

Page 173: Semi-automatic history matching applying parameter ... · Semi-automatic history matching. applying . parameter estimation techniques. ... • History matching ... Workflow. 8. facies.

Smart Fields, Stanford UniversityIntegrated Operations 200707

Preliminary Results

• 20x20x10 sector from zone 3

• m: facies

• k, φ: regression from well location

• VP and VS as seismics

• PCA: 30 coefficients retained– 1000 realizations

21

(joint work with Mukerji and Santos)

Page 174: Semi-automatic history matching applying parameter ... · Semi-automatic history matching. applying . parameter estimation techniques. ... • History matching ... Workflow. 8. facies.

Smart Fields, Stanford UniversityIntegrated Operations 200707

Preliminary Results

• 20x20x10 sector from zone 3

• m: facies

• k, φ: regression from well location

• VP and VS as seismics

• PCA: 30 coefficients retained

• Measured data only after 3 months

21

(joint work with Mukerji and Santos)

Page 175: Semi-automatic history matching applying parameter ... · Semi-automatic history matching. applying . parameter estimation techniques. ... • History matching ... Workflow. 8. facies.

Smart Fields, Stanford UniversityIntegrated Operations 200707

22

Preliminary Results(joint work with Mukerji and Santos)

Page 176: Semi-automatic history matching applying parameter ... · Semi-automatic history matching. applying . parameter estimation techniques. ... • History matching ... Workflow. 8. facies.

Smart Fields, Stanford UniversityIntegrated Operations 200707

• Alternating optimization strategy– production ( 2 iter)

– production + seismics ( 2 iter)

– seismics (10 iter)

22

Preliminary Results(joint work with Mukerji and Santos)

Page 177: Semi-automatic history matching applying parameter ... · Semi-automatic history matching. applying . parameter estimation techniques. ... • History matching ... Workflow. 8. facies.

Smart Fields, Stanford UniversityIntegrated Operations 200707

• Alternating optimization strategy

22

Preliminary Results(joint work with Mukerji and Santos)

Page 178: Semi-automatic history matching applying parameter ... · Semi-automatic history matching. applying . parameter estimation techniques. ... • History matching ... Workflow. 8. facies.

Smart Fields, Stanford UniversityIntegrated Operations 200707

• Alternating optimization strategy

• Optimizer: SQP + numerical gradients

22

Preliminary Results(joint work with Mukerji and Santos)

Page 179: Semi-automatic history matching applying parameter ... · Semi-automatic history matching. applying . parameter estimation techniques. ... • History matching ... Workflow. 8. facies.

Smart Fields, Stanford UniversityIntegrated Operations 200707

• Alternating optimization strategy

• Optimizer: SQP + numerical gradients

• True model m*: original sector projected over truncated PCA basis

*m

22

Preliminary Results(joint work with Mukerji and Santos)

Page 180: Semi-automatic history matching applying parameter ... · Semi-automatic history matching. applying . parameter estimation techniques. ... • History matching ... Workflow. 8. facies.

Smart Fields, Stanford UniversityIntegrated Operations 200707

• Alternating optimization strategy

• Optimizer: SQP + numerical gradients

• True model m*: original sector projected over truncated PCA basis

• Initial guess m0*: random realization

taken from the 1000 for the PCA

*m

22

Preliminary Results(joint work with Mukerji and Santos)

Page 181: Semi-automatic history matching applying parameter ... · Semi-automatic history matching. applying . parameter estimation techniques. ... • History matching ... Workflow. 8. facies.

Smart Fields, Stanford UniversityIntegrated Operations 200707

Production Data

23

(joint work with Mukerji and Santos)

Page 182: Semi-automatic history matching applying parameter ... · Semi-automatic history matching. applying . parameter estimation techniques. ... • History matching ... Workflow. 8. facies.

Smart Fields, Stanford UniversityIntegrated Operations 200707

Production Data

5 spot

injector

producer

23

(joint work with Mukerji and Santos)

Page 183: Semi-automatic history matching applying parameter ... · Semi-automatic history matching. applying . parameter estimation techniques. ... • History matching ... Workflow. 8. facies.

Smart Fields, Stanford UniversityIntegrated Operations 200707

Oil Production

23

(joint work with Mukerji and Santos)

Page 184: Semi-automatic history matching applying parameter ... · Semi-automatic history matching. applying . parameter estimation techniques. ... • History matching ... Workflow. 8. facies.

Smart Fields, Stanford UniversityIntegrated Operations 200707

Oil Production

23

(joint work with Mukerji and Santos)

Page 185: Semi-automatic history matching applying parameter ... · Semi-automatic history matching. applying . parameter estimation techniques. ... • History matching ... Workflow. 8. facies.

Smart Fields, Stanford UniversityIntegrated Operations 200707

Oil Production

23

(joint work with Mukerji and Santos)

Page 186: Semi-automatic history matching applying parameter ... · Semi-automatic history matching. applying . parameter estimation techniques. ... • History matching ... Workflow. 8. facies.

Smart Fields, Stanford UniversityIntegrated Operations 200707

Oil Production

23

(joint work with Mukerji and Santos)

Page 187: Semi-automatic history matching applying parameter ... · Semi-automatic history matching. applying . parameter estimation techniques. ... • History matching ... Workflow. 8. facies.

Smart Fields, Stanford UniversityIntegrated Operations 200707

Water Injection

23

(joint work with Mukerji and Santos)

Page 188: Semi-automatic history matching applying parameter ... · Semi-automatic history matching. applying . parameter estimation techniques. ... • History matching ... Workflow. 8. facies.

Smart Fields, Stanford UniversityIntegrated Operations 200707

Water Injection

23

(joint work with Mukerji and Santos)

Page 189: Semi-automatic history matching applying parameter ... · Semi-automatic history matching. applying . parameter estimation techniques. ... • History matching ... Workflow. 8. facies.

Smart Fields, Stanford UniversityIntegrated Operations 200707

Water Injection

23

(joint work with Mukerji and Santos)

Page 190: Semi-automatic history matching applying parameter ... · Semi-automatic history matching. applying . parameter estimation techniques. ... • History matching ... Workflow. 8. facies.

Smart Fields, Stanford UniversityIntegrated Operations 200707

Water Injection

23

(joint work with Mukerji and Santos)

Page 191: Semi-automatic history matching applying parameter ... · Semi-automatic history matching. applying . parameter estimation techniques. ... • History matching ... Workflow. 8. facies.

Smart Fields, Stanford UniversityIntegrated Operations 200707

Seismic Data

5 spot

injector

producer

24

(joint work with Mukerji and Santos)

Page 192: Semi-automatic history matching applying parameter ... · Semi-automatic history matching. applying . parameter estimation techniques. ... • History matching ... Workflow. 8. facies.

Smart Fields, Stanford UniversityIntegrated Operations 200707

5 spot

injector

producer

24

(joint work with Mukerji and Santos)

Seismic Data

Page 193: Semi-automatic history matching applying parameter ... · Semi-automatic history matching. applying . parameter estimation techniques. ... • History matching ... Workflow. 8. facies.

Smart Fields, Stanford UniversityIntegrated Operations 200707

Velocities (VP )

true(projected)

initialguess

matchingproduction

(2 iter)

alternatingmatching

24

(joint work with Mukerji and Santos)

Page 194: Semi-automatic history matching applying parameter ... · Semi-automatic history matching. applying . parameter estimation techniques. ... • History matching ... Workflow. 8. facies.

Smart Fields, Stanford UniversityIntegrated Operations 200707

Velocities (VP )

true(projected)

initialguess

matchingproduction

(2 iter)

alternatingmatching

24

(joint work with Mukerji and Santos)

Page 195: Semi-automatic history matching applying parameter ... · Semi-automatic history matching. applying . parameter estimation techniques. ... • History matching ... Workflow. 8. facies.

Smart Fields, Stanford UniversityIntegrated Operations 200707

Velocities (VP )

initialguess

matchingproduction

(2 iter)

alternatingmatching

24

(joint work with Mukerji and Santos)

Page 196: Semi-automatic history matching applying parameter ... · Semi-automatic history matching. applying . parameter estimation techniques. ... • History matching ... Workflow. 8. facies.

Smart Fields, Stanford UniversityIntegrated Operations 200707

Velocities (VP )

x

z

y

z

24

true

after production

initial

alternating

(joint work with Mukerji and Santos)

Page 197: Semi-automatic history matching applying parameter ... · Semi-automatic history matching. applying . parameter estimation techniques. ... • History matching ... Workflow. 8. facies.

Smart Fields, Stanford UniversityIntegrated Operations 200707

Facies

25

(joint work with Mukerji and Santos)

Page 198: Semi-automatic history matching applying parameter ... · Semi-automatic history matching. applying . parameter estimation techniques. ... • History matching ... Workflow. 8. facies.

Smart Fields, Stanford UniversityIntegrated Operations 200707

Facies

true(projected)

initialguess

matchingproduction

(2 iter)

alternatingmatching

25

(joint work with Mukerji and Santos)

Page 199: Semi-automatic history matching applying parameter ... · Semi-automatic history matching. applying . parameter estimation techniques. ... • History matching ... Workflow. 8. facies.

Smart Fields, Stanford UniversityIntegrated Operations 200707

Facies

x

y

LAYER 125

true

after production

initial

alternating

(joint work with Mukerji and Santos)

Page 200: Semi-automatic history matching applying parameter ... · Semi-automatic history matching. applying . parameter estimation techniques. ... • History matching ... Workflow. 8. facies.

Smart Fields, Stanford UniversityIntegrated Operations 200707

Facies

x

y

LAYER 125

true

after production

initial

alternating

(joint work with Mukerji and Santos)

Page 201: Semi-automatic history matching applying parameter ... · Semi-automatic history matching. applying . parameter estimation techniques. ... • History matching ... Workflow. 8. facies.

Smart Fields, Stanford UniversityIntegrated Operations 200707

Facies

x

y

LAYER 425

true

after production

initial

alternating

(joint work with Mukerji and Santos)

Page 202: Semi-automatic history matching applying parameter ... · Semi-automatic history matching. applying . parameter estimation techniques. ... • History matching ... Workflow. 8. facies.

Smart Fields, Stanford UniversityIntegrated Operations 200707

Facies

x

y

LAYER 425

true

after production

initial

alternating

(joint work with Mukerji and Santos)

Page 203: Semi-automatic history matching applying parameter ... · Semi-automatic history matching. applying . parameter estimation techniques. ... • History matching ... Workflow. 8. facies.

Smart Fields, Stanford UniversityIntegrated Operations 200707

Facies

x

y

LAYER 625

true

after production

initial

alternating

(joint work with Mukerji and Santos)

Page 204: Semi-automatic history matching applying parameter ... · Semi-automatic history matching. applying . parameter estimation techniques. ... • History matching ... Workflow. 8. facies.

Smart Fields, Stanford UniversityIntegrated Operations 200707

Facies

x

y

LAYER 625

true

after production

initial

alternating

(joint work with Mukerji and Santos)

Page 205: Semi-automatic history matching applying parameter ... · Semi-automatic history matching. applying . parameter estimation techniques. ... • History matching ... Workflow. 8. facies.

Smart Fields, Stanford UniversityIntegrated Operations 200707

Facies

x

y

LAYER 1025

true

after production

initial

alternating

(joint work with Mukerji and Santos)

Page 206: Semi-automatic history matching applying parameter ... · Semi-automatic history matching. applying . parameter estimation techniques. ... • History matching ... Workflow. 8. facies.

Smart Fields, Stanford UniversityIntegrated Operations 200707

Facies

x

y

LAYER 1025

true

after production

initial

alternating

(joint work with Mukerji and Santos)

Page 207: Semi-automatic history matching applying parameter ... · Semi-automatic history matching. applying . parameter estimation techniques. ... • History matching ... Workflow. 8. facies.

Smart Fields, Stanford UniversityIntegrated Operations 200707

Outline

O

• Smart Fields Consortium

• The History Matching Problem

• Challenges and Trends

• History Matching at Smart Fields

• Conclusions

Page 208: Semi-automatic history matching applying parameter ... · Semi-automatic history matching. applying . parameter estimation techniques. ... • History matching ... Workflow. 8. facies.

Smart Fields, Stanford UniversityIntegrated Operations 200707

Conclusions

• Semi-automatic history matching

• Seen as an optimization problem

• Local vs. global optimization

• KPCA reduces number of optimization variables and honors geology

• Different types of data integrated

C

Page 209: Semi-automatic history matching applying parameter ... · Semi-automatic history matching. applying . parameter estimation techniques. ... • History matching ... Workflow. 8. facies.

Smart Fields, Stanford UniversityIntegrated Operations 200707

Acknowledgement

• Khalid Aziz, Jef Caers, Lou Durlofsky, Roland Horne, Tapan Mukerji and Eduardo Santos

• Marco Cardoso, Jerome Onwunalu, Danny Rojas and Pallav Sarma

• Others at the Smart Field Consortium

A

Page 210: Semi-automatic history matching applying parameter ... · Semi-automatic history matching. applying . parameter estimation techniques. ... • History matching ... Workflow. 8. facies.

Smart Fields, Stanford UniversityIntegrated Operations 200707

Thank youfor

your attention!

October 3, 2007

D. EcheverrD. Echeverríía Ciaurria Ciaurri

Smart Fields ConsortiumStanford University

Page 211: Semi-automatic history matching applying parameter ... · Semi-automatic history matching. applying . parameter estimation techniques. ... • History matching ... Workflow. 8. facies.

Smart Fields, Stanford UniversityIntegrated Operations 200707

Semi-automatic history matchingapplying

parameter estimation techniques

October 3, 2007

D. EcheverrD. Echeverríía Ciaurria Ciaurri

Smart Fields ConsortiumStanford University