EDGER

Department of Geological SciencesThe University of Texas at Austin

Mrinal K. Sen & Kyle T. Spikes

Dissemination  of  ResultsParticipants in the Forum are invited to attend the Annual ResearchSymposium each spring, which focuses on research results. Members arewelcome to provide input of specific industry research and/or educationalneeds. Results will also be shared with sponsors through personal on-sitevisits, preprints of research papers, compilations of educational activities andresults, and research computer codes from individual projects. For moreinformation, please contact:

Mrinal Sen & Kyle SpikesForum Coordinators

The University of Texas at AustinJackson School of Geosciences

Dept. of Geological Sciences2275 Speedway Stop C9000

Austin, TX 78712 – 1722

2

Currently  Available  Software  developed  by  students

• Basis pursuit inversion (BPI)• Greedy annealed importance

sampling (GAIS)• Rock physics modeling for

unconventional reservoirs• Bayesian rock physics analysis

3

Pictorial representation of Kernel MatrixWhere: - Kernel Matrix

- Even Seismic response- Odd Seismic response

BPI  -­ Reflectivity  Inversion

NN

NΘΨ=Φ

Theory Reflectivity Inversion

4

BPI  -­ Inverted  Reflection  Coefficients

5

Pre-stack seismic Inverted RpInverted Rs

Greedy  Annealed  Importance  Sampling  (GAIS)

6

ROCK-­PHYSICS  MODELING  AND  GRID  SEARCHING  WORKFLOW

Modeled VP & VS (or   IP &  IS)

VP & VS(or  IP &  IS from Seismic Data

Grid  Searching

Estimated  Reservoir  Properties

Background  Matrix

Self-­‐Consistent  Model

Chapman’s  Model

Effective  Shale

QuartzCalcite

Pyrite

Non-­‐Spherical  Pore

Kerogen

Clay

Spherical  Pore

Crack

Aligned  Fracture

Rock  Matrix

1) Input minerals and pores in the self consistent model

to form the background matrix2) Imbed spherical pores, cracks, and aligned

fracturesusing Chapman’s 2002 model

3) Obtain an effective shale model4) Use grid searching to estimate reservoir

propertiesfrom log- or seismic-scale elastic properties

1)

2)

3)

4)

7

THREE-­STEP  DEM  METHOD

Step  1Input  stiff  inclusions  (round)and  soft  (ellipsoidal)  inclusions)

Step  2  Re-­orient  soft  inclusions  so  that  they  are  not  aligned,  thus  reducing  modeled  anisotropy

Step  3Use  Vp/Vs  relationships  to  account  for  effective  pressureCal

Qua

Clay

Ker

Cal

Cal

Clay

Clay

Pyr

Dolo

Pore

Pore

Pore

Clay20 40 60 80 100

10

15

20

25

30

Effective Pressure (MPa)

S−w

ave

Mod

ulus

(GPa

)Stiffness  (GPa)

C66

C44

20 40 60 80 10030

40

50

60

70

80

Effective Pressure (MPa)

P−w

ave

Mod

ulus

(GPa

)Stiffness  (GPa)

C11

C33

8

In  Progress• Modeling of seismic response of

discrete fracture networks based on integral formulation.

• Quantum annealing (QA) inversion of angle gathers.

• Probabilistic rock physics templates.• Combined Biot and squirt-flow

modeling for sonic wave modes.• Neural-network based image

segmentation.9

Thanks  to  our  sponsors

10