www.oplnk.net/~glinsky

Integration of uncertain subsurface information into multiple reservoir simulation models

Michael E. Glinsky, Bruce Asher, Robin Hill, Mark Flynn, and Mark Stanley, BHP Billiton PetroleumJames Gunning, CSIRO PetroleumTroy Thompson, DownUnder GeoSolutionsJerome Kalifa, and Stephane Mallat, Ecole PolytechniqueDidier Renard, Ecole des MinesChris White, LSU

Integration of uncertain subsurface information into multiple reservoir simulation models4 April 2005Page 2

Outline

• Description of Stybarrow oil field• Use of sparse spike inversion and the correlation wavelet• Probabilistic wavelet derivation giving critical noise level• Probabilistic model based inversion with uncertainty• “Massaging” the results into the reservoir simulation model• “Decoration” and “Enforcement”

– adding the subseismic structure

Integration of uncertain subsurface information into multiple reservoir simulation models4 April 2005Page 3

Location of Stybarrow field

WESTERNAUSTRALIA

ExmouthPlateau

BarrowSub-basin

Exmou

th

Sub -ba

sin

InvestigatorSub-Basin

Cape Range Fracture Zone

NovaraArch

Reso

lution

Arc

h

Kanga

roo Syn

cline

Exm

outh

Pla

teau

Arc

h Oil Field

Gas Field

Oil & Gas Field

Jurassic to EarlyCretaceousDepocentre

Early CretaceousDepocentre

Triassic-Cored Structural Highs

Ningaloo ArchWA-255-P

50 km

Barrow Island

Scarborough

MacedonStybarrow

Integration of uncertain subsurface information into multiple reservoir simulation models4 April 2005Page 4

Structure map of Stybarrow field

Depth Structure Map of the Top Macedon Member

Stybarrow Oil Field

Laverda Oil and Gas Field

WA-271-P

WA-255-P(1)

1

2

34

N

Integration of uncertain subsurface information into multiple reservoir simulation models4 April 2005Page 5

Geologic cross section of Stybarrow

ReservoirDingo Fm Source Rock

Seal

NW SE

Integration of uncertain subsurface information into multiple reservoir simulation models4 April 2005Page 6

Seismic data and sparse spike inversion with secant amplitude extraction

soft

hard

Impedance

2.16 s

2.40 s

Stybarrow-3Stybarrow-4

Stybarrow-2 Stybarrow-1

-10%

+10%

Integration of uncertain subsurface information into multiple reservoir simulation models4 April 2005Page 7

Definition of secant amplitude and area

impedance vs. time

A

area

WP1 P3 P2

net sand = secant area * Vsand / 2 Rsand

Integration of uncertain subsurface information into multiple reservoir simulation models4 April 2005Page 8

0

5

10

15

20

25

30

35

Stybarrow-1

Stybarrow-2

Stybarrow-3 Stybarrow-4

Net sand calculated from secant area

net sand (m)

Integration of uncertain subsurface information into multiple reservoir simulation models4 April 2005Page 9

Geometry used for inversion benchmark

400 ft

station0 200

8500 ft/s2.33 gm/cc68% N/G 9434 ft/s

2.37 gm/cc

7134 ft/s2.01 gm/cc

Integration of uncertain subsurface information into multiple reservoir simulation models4 April 2005Page 10

Inversion benchmark

0

100

200

300

400

0 50 100 150 200

station

net

san

d (

ft)

uninverted dataindustry inversion #1industry inversion #2new inversiontrue value

Integration of uncertain subsurface information into multiple reservoir simulation models4 April 2005Page 11

Stybarrow-1

Stybarrow-2

Stybarrow-3 Stybarrow-4

Oil probability map calculated from secant amplitude

0

0.1

0.2

0.3

0.4

0.5

0.6

0.7

0.8

0.9

1

shale

oilgas

brine

0%

-8%

-16%

RFC

Stybarrow-1

Stybarrow-2

Stybarrow-3 Stybarrow-4

1.24 1.25 1.26 1.270%

-8%

-16%

secant amplitude

Integration of uncertain subsurface information into multiple reservoir simulation models4 April 2005Page 12

Tie of seismic data to synthetic seismic using probabilistic wavelet derivation

2.0 s

2.4 s

timegreen = seismicred = synthetic

Integration of uncertain subsurface information into multiple reservoir simulation models4 April 2005Page 13

Most likely probabilistic wavelet compared to correlation wavelet

probabilisticcorrelation

-60 ms

0 ms

60 ms

note: probabilistic wavelet is much shorter than correlation wavelet

Integration of uncertain subsurface information into multiple reservoir simulation models4 April 2005Page 14

Most likely probabilistic wavelet compared to ensemble of possible wavelets

-20 ms

20 ms

0 ms

note: noise is 17% size of oil reflector, SNR is 15 dB

Integration of uncertain subsurface information into multiple reservoir simulation models4 April 2005Page 15

Probabilistic model based inversion

• Layer based model built at seismic loop scale using sparse spikeinversion

• Standard rock physics correlations estimated with uncertainty• Fundamental properties of layers are:

– net-to-gross ratio (N/G)– layer top and base– fluid type

• Ensemble of models generated that are consistent with seismic towithin estimated noise level

Integration of uncertain subsurface information into multiple reservoir simulation models4 April 2005Page 16

Ensemble of models at Stybarrow-1 well location show effect of model based inversion

soft

hard

acoustic impedance

before after

realization realization

2.2 s

2.4 s

Integration of uncertain subsurface information into multiple reservoir simulation models4 April 2005Page 17

Effect of model based inversion on match of synthetic seismic to seismic data

2.24 s

2.36 s

before after seismic

Integration of uncertain subsurface information into multiple reservoir simulation models4 April 2005Page 18

Inversion tightens the range of possible net sand

before

after

well value0 m 50 m

probability of oil increased to 97% from 50% (oil in sand at this location)

Integration of uncertain subsurface information into multiple reservoir simulation models4 April 2005Page 19

Effect of inversion on estimation of N/G in main hydrocarbon (krieged to wells)

N/G0%

100%

STD N/G

before

after

after

before0%

30%

Integration of uncertain subsurface information into multiple reservoir simulation models4 April 2005Page 20

Estimation of oil probability

well control only inversion only0%

100%

Integration of uncertain subsurface information into multiple reservoir simulation models4 April 2005Page 21

Cross sections of mean model

before

after

2.40 s

2.16 s0%

100%

Integration of uncertain subsurface information into multiple reservoir simulation models4 April 2005Page 22

Net sand prediction ahead of drill bit

7.612.3 ± 4.3Stybarrow -4

29.1 ± 6.4Stybarrow-3

result (m)prediction (m)well

40% probability well values larger deviation than observed

Integration of uncertain subsurface information into multiple reservoir simulation models4 April 2005Page 23

Results of model based inversion “massaged” onto reservoir model

• Lateral correlation added• Short range noise eliminated• Put onto irregular corner point geometry• Fault blocks honored• Inter-layer and inter-property correlation honored

Integration of uncertain subsurface information into multiple reservoir simulation models4 April 2005Page 24

Difference between the seismic grid and the reservoir simulation grid

Stybarrow-1

1400

1600

1800

2000

2200

2400

2600

2800

Stybarrow-1

seismic grid reservoir simulation grid

1.2 km x 1.2 km

Integration of uncertain subsurface information into multiple reservoir simulation models4 April 2005Page 25

0

5

10

15

20

25

30

35

Stybarrow-1

Stybarrow-2

Stybarrow-3 Stybarrow-4

Stybarrow-1

Stybarrow-2

Stybarrow-3 Stybarrow-4

Effect of massaging on mean net sand map

after

net sand (m)

7 km x 7 km

before

Integration of uncertain subsurface information into multiple reservoir simulation models4 April 2005Page 26

Stybarrow-1

Stybarrow-2

Stybarrow-3 Stybarrow-4

Stybarrow-1

Stybarrow-2

Stybarrow-3 Stybarrow-4

Stybarrow-1

Stybarrow-2

Stybarrow-3 Stybarrow-4

Stybarrow-1

Stybarrow-2

Stybarrow-3 Stybarrow-4

Stybarrow-1

Stybarrow-2

Stybarrow-3 Stybarrow-4

Realizations of net sand generated by massager

net sand (m)

0

35

Integration of uncertain subsurface information into multiple reservoir simulation models4 April 2005Page 27

-2180

-2160

-2140

-2120

-2100

-2080

-2060

-2040

-2180

-2160

-2140

-2120

-2100

-2080

-2060

-2040

-2180

-2160

-2140

-2120

-2100

-2080

-2060

-2040

Cross sections through model realizations2040 m

2180 m

N/G

0%

100%

Stybarrow-1

Stybarrow-2

2.6 km

well

fault

note: models in depth

Integration of uncertain subsurface information into multiple reservoir simulation models4 April 2005Page 28

-2180

-2160

-2140

-2120

-2100

-2080

-2060

-2040

-2180

-2160

-2140

-2120

-2100

-2080

-2060

-2040

-2180

-2160

-2140

-2120

-2100

-2080

-2060

-2040

-2180

-2160

-2140

-2120

-2100

-2080

-2060

-2040

-2180

-2160

-2140

-2120

-2100

-2080

-2060

-2040

-2180

-2160

-2140

-2120

-2100

-2080

-2060

-2040

Reservoir simulation models are “decorated with subseismic structure

before decorationafter proportional decoration

after offlapping decoration

N/G

0%

100%

Integration of uncertain subsurface information into multiple reservoir simulation models4 April 2005Page 29

Conclusions

• Ensemble of reservoir simulation models consistent with:– Seismic data– Well information

– Geologic concepts

• Gives:– Volumetric distributions– Minimum net sand (well completion)– Etc.

• Range of production profiles– Potential production– Risks for development