ANNUAL MEETING MASTER OF PETROLEUM ENGINEERING

The CERENA-IV synthetic dataset

28/May/2014 Instituto Superior Técnico

Performance assessment between

deterministic and geostatistical

seismic inversion methodologies.

1

Table of contents

1. Overview & Objectives

2. Methods

Deterministic approaches: Model Based Inversion; Sparse Spike

Inversion

Stochastic approaches: Global Stochastic Inversion; Global Elastic

Inversion; Geostatistical Inversion of Seismic AVO data directly to facies

models

3. Work in Progress & Preliminary results

4. To do next…

28/May/2014 Instituto Superior Técnico 2

Overview & Objectives

28/May/2014 Instituto Superior Técnico 3

• Geophysical inverse problems infer

subsurface properties from a set of indirect

measurements:

d=F(m)+e

• Deterministic

• Bayesian inference frameworks are natural

ways to tackle geophysical inverse problems.

– Stochastic or geostatistical

Overview & Objectives

28/May/2014 Instituto Superior Técnico 4

• Synthetic dimensions:

– Grid size: 121x200x250

– 6 050 000 cells

• 3 normal faults

• 4 facies groups

• Constrained dataset:

– 14 wells placed randomly

– Seismic (post and pre stack)

Facies Seismic

rho Vp Vs

Well logs

Deterministic approaches

28/May/2014 Instituto Superior Técnico 5

(Adapted from Russel, 1988)

Sparse Spike Inversion (Linear Programming)

• Puts events only where the

seismic demands

• Attempts to produce the

simplest model consistent with

the seismic data

• It often produces fewer events

that are known to be true

Deterministic approaches

28/May/2014 Instituto Superior Técnico 6

Model Based Inversion

(Adapted from Russel, 1988)

• Produces a broadband, high

frequency result

• The results can be highly

dependent on the initial guess

model: filtering the model may

lessen its effect

• The effective resolution of the

seismic is enhanced

• There is a non uniqueness

problem, as with all inversion

Global stochastic approaches

28/May/2014 Instituto Superior Técnico 7

• Iterative methodologies

• Model perturbation recurs to sequential simulation and co-simulation

• Global optimizer: genetic algorithm

(Adapted from Azevedo, 2013)

Global Stochastic Inversion

28/May/2014 Instituto Superior Técnico 8

(Adapted from Azevedo, 2013)

Global Elastic Inversion

28/May/2014 Instituto Superior Técnico 9

(Adapted from Azevedo, 2013)

Geostatistical Inversion of seismic

AVO data directly to facies models

28/May/2014 Instituto Superior Técnico 10

(Adapted from Azevedo, 2013)

Work in Progress & Preliminary

Results

28/May/2014 Instituto Superior Técnico 11

Parameters used in the applied

methodologies

• Global Stochastic Inversion

• Well data, variograms

• 32 cubes

• 6 iterations

• LP Sparse Spike

• 10% sparseness

• 8Hz constraint

• Window length: 128 samples

• Model Based

• Low frequency model w/ low pass filter

5-12Hz

• 10 iterations

• Available data:

• Well logs (AI, Vp, Vs, Rho, Facies)

• Seismic data (pre and post-stack)

• wavelet

• Control data

• Original model

rho

Vp

Vs

AI

Work in Progress & Preliminary

Results – seismic assessment

28/May/2014 Instituto Superior Técnico 12

Global Stochastic Inversion Original Model Arithmetic mean AI_DSS_6_4 AI_DSS_6_14

Sparse Spike Model Based Original Model Sparse Spike

Model Based

GSI (AI_DSS_6_4)

Original Model

28/May/2014 Instituto Superior Técnico 13

Sparse Spike

Model Based

GSI (AI_DSS_6_4)

Original Model

Synthetic Seismic vs. Real Model Seismic

Inversion Method Correlation Coefficient

LP Sparse Spike 0.9691

Model Based 0.9854

GSI 0.9120

Work in Progress & Preliminary

Results – seismic assessment

Work in Progress & Preliminary

Results – AI assessment

28/May/2014 Instituto Superior Técnico 14

Sparse Spike Model Based Original Model

GSI_Variance GSI_Arithmetic mean GSI_AI_DSS_6_4 GSI_AI_DSS_6_14

Model Based Sparse Spike

GSI_AI_DSS_6_4

To do next…

28/May/2014 Instituto Superior Técnico 15

1. Apply the remaining methods to the case study

2. Fine tune the models to reach the intended correlations that

will permit assess the performance of the applied

methodologies

Thank you.

References

28/May/2014 Instituto Superior Técnico 16

• Azevedo, L. (2013). Geostatistical methods for integrating seismic reflection data into

subsurface Earth models. Lisboa: Instituto Superior Técnico.

• Caetano, H. (2009). Integration of Seismic Information in Reservoir Models: Global

Stochastic Inversion. Lisboa: Instituto Superior Técnico.

• Cooke, D., & Schneider, A. (1983). Generalized linear inversion of reflection seismic data.

GEOPHYSICS. VOL. 48. NO 6, pp. 665-676.

• Nunes, R., Soares, A., & al, e. (2012). Geostatistical Inversion of Prestack Seismic Data.

• Oldenburg, D., Scheuer, T., & Levy, S. (1983). Recovery of the acoustic impedance from

reflection seismograms. GEOPHYSICS, VOL. 48, NO. 10, pp. 1318-1337.

• Russel, B. (1988). Introduction to Seismic Inversion Methods. Calgary, Alberta: Society of

Exploration Geophysicists.

• Russell, B., & Hampson, D. (1991). Comparison of Poststack Seismic Inversion Methods.

61st Annual International Meeting - SEG, pp. 876-878.