FORCE seminar - 06/02/02. Prestack Data. NPD Stavanger.
Philippe JULIEN - Massive modelling and AVO
FORCE: Prestack Data - The key to modern interpretation
THE IMPORTANCE OF PRESTACK SEISMIC
MASSIVE MODELLING FOR A.V.O. CALIBRATION AND
SEISMIC RESERVOIR CHARACTERIZATION
Authors: Ph. Julien (speaker), F. Pivot Acknowledgements: A. Douillard, J. Marrauld, C. Borghi
FORCE seminar - 06/02/02. Prestack Data. NPD Stavanger.
Philippe JULIEN - Massive modelling and AVO
INTRODUCTION
* Seismic reservoir characterization more andmore uses prestack data.
* We will focus on the A.V.O. (Amplitude VersusOffset) calibration & prediction.
FORCE seminar - 06/02/02. Prestack Data. NPD Stavanger.
Philippe JULIEN - Massive modelling and AVO
Now
ComputeComputeseismicseismic
attributesattributes
ComputeComputeseismicseismic
attributesattributes
CalibrationCalibration??
3Dseismic data
3Dreservoir prop.
nDseismic data
3Dreservoir prop.
Before...
The calibration problem in seismic reservoir characterizationThe calibration problem in seismic reservoir characterization
CalibrationCalibration?????????????? 3D
uncertainties.
FORCE seminar - 06/02/02. Prestack Data. NPD Stavanger.
Philippe JULIEN - Massive modelling and AVO
OVERVIEW of QUESTIONS in the frame of the LITHOSEISMIC project
SEISMIC WORLDSEISMIC WORLD
GEOLOGICAL WORLDGEOLOGICAL WORLD
* What is the sensitivity of the seismic* What is the sensitivity of the seismicresponse for a given geological variation?response for a given geological variation?
* Which geological property can be * Which geological property can be mapped with seismic (and which onemapped with seismic (and which one can not) ? can not) ?
* Which seismic attributes for* Which seismic attributes for solving a geological problem? solving a geological problem? * Which geological constraints must be* Which geological constraints must beadded?added? * Is the seismic response unique?* Is the seismic response unique?
* Which uncertainties* Which uncertainties??
(tectonics, sedimentology, petrophysics)
==>3 steps approach
FORCE seminar - 06/02/02. Prestack Data. NPD Stavanger.
Philippe JULIEN - Massive modelling and AVO
WHY SO MANY AVO FAILURES?
A.V.O is an efficient method for fluidscharacterization… but too many failures in the past.
1) AVO depends on fluids (and their saturations).
FORCE seminar - 06/02/02. Prestack Data. NPD Stavanger.
Philippe JULIEN - Massive modelling and AVO
WHY SO MANY AVO FAILURES?
1) AVO depends on fluids (and their saturations).
2) But AVO also depends on:
* lithology Ex.: - coal layers; - Vclay.
FORCE seminar - 06/02/02. Prestack Data. NPD Stavanger.
Philippe JULIEN - Massive modelling and AVO
WHY SO MANY AVO FAILURES?
1) AVO depends on fluids (and their saturations).
2) But AVO also depends on:
* lithology: Ex.: - coal layers; - Vclay.
* reservoir thickness: Ideally, AVO is reliable forreservoirs with a thickness larger than the tuningthickness.
Note: Amplitudes depend on porosity.
FORCE seminar - 06/02/02. Prestack Data. NPD Stavanger.
Philippe JULIEN - Massive modelling and AVO
IMPLICATIONS
* AVO inversion does not lead (in general) to a uniquesolution.
FORCE seminar - 06/02/02. Prestack Data. NPD Stavanger.
Philippe JULIEN - Massive modelling and AVO
IMPLICATIONS
* AVO inversion does not lead (in general) to a uniquesolution.
* Usually, some seismic modelling (~ 10 modellingexperiments) is carried out to understand the AVOphenomenom.
* This is not enough: hundreds of seismic modellingexperiments are mandatory to begin understandingthe seismic response…
FORCE seminar - 06/02/02. Prestack Data. NPD Stavanger.
Philippe JULIEN - Massive modelling and AVO
AVO analysis with massive seismic AVO analysis with massive seismic modellingmodellingEXAMPLE OF APPLICATIONEXAMPLE OF APPLICATION Example : Post-Mortem StudyExample : Post-Mortem Study
All models with phi below 20%,No tuning (overburden shale layer and reservoir >150 m)
Gas(Sw20/60%)
Brine
• Fluid discrimination difficult as brine case (phi=10%) is very similar to gas response (phi=15%). May be possible on long offsets.
Impact of fluid content on AVO effectImpact of fluid content on AVO effect
FORCE seminar - 06/02/02. Prestack Data. NPD Stavanger.
Philippe JULIEN - Massive modelling and AVO
MASSIVE SEISMIC MODELLING: a mandatory step
* Hundreds of seismic modelling experiments aremandatory to begin understanding the seismicresponse…
* References: - O. Youn (Unocal) & H. Zhou (Exxon), SEG 98; - F. Pivot et al. (TotalFinaElf), EAGE 99; - F. Pivot & O. Balz (TotalFinaElf), EAGE 00; - Y. El Ouair & F. Pivot (TotalFinaElf), SEG 01.
* Such calibration & modelling stages will have an impact onseismic acquisition, processing and interpretation.
FORCE seminar - 06/02/02. Prestack Data. NPD Stavanger.
Philippe JULIEN - Massive modelling and AVO
OVERVIEW OF THE PRESENTATION
I) A few recalls: Petro-elastic modelling, amplitudecalibration and 1D seismic massive modelling.
II) Application: A real data example: impact onacquisition, processing and interpretation.
III) 2D or 3D perturbations: Real examples; impacts onthe interpretation.
FORCE seminar - 06/02/02. Prestack Data. NPD Stavanger.
Philippe JULIEN - Massive modelling and AVO
A.V.O. CALIBRATION: the general processA.V.O. CALIBRATION: the general process
GoalCalibration: to find a relationship between well and seismic data.Prediction: to extrapolate this relationship to the whole seismic cubes.
PETRO-ELASTIC MODELLING, AMPLITUDE CALIBRATIONand MASSIVE PRESTACK MODELLING.
Petrophysical measurements
Elastic measurements/logs
1:Petro-elastic modellingSurface seismics
Well seismics
2: Amplitude calibration
(Massive) prestack models
2/3:Seismic modelling 4:Impacts
FORCE seminar - 06/02/02. Prestack Data. NPD Stavanger.
Philippe JULIEN - Massive modelling and AVO
Synergy for calibration: the petro-elastic model (1st step)AmplitudesAmplitudes
DensityDensityP SonicP Sonic
AVOAVO
S SonicS Sonic
CoresCores
LogsLogs
SedimentologySedimentology
PetroPetro-elastic-elasticModelModel
PETROPHYSICAL PARAM.E
LA
STIC
PA
RA
M.
FORCE seminar - 06/02/02. Prestack Data. NPD Stavanger.
Philippe JULIEN - Massive modelling and AVO
Second step: AVO CALIBRATIONMULTI-TIE & AVA TIE
NEAR FARw
ater
gas
IP POISSON GR Log VSP Seis south east north west Walkaway
Log Seis south east north west Walkaway
MU
LTI-T
IEA
VA T
IE
Surface seismic coherentwith walkaway
AM
PLIT
UD
E
ANGLE (°)
ANALYSIS AT 2132 ms
logsurface seismic
walkaway southwalkaway northwalkaway eastwalkaway west
FORCE seminar - 06/02/02. Prestack Data. NPD Stavanger.
Philippe JULIEN - Massive modelling and AVO
GENERALGENERAL CONCEPT OF MASSIVE MODELLING (3rd step)CONCEPT OF MASSIVE MODELLING (3rd step)
Petro & Structuralperturbations
Inputmodel
ρρρρ,Vp,Vs Pert. blockymodels
(ρρρρ,Vp,Vs)new triplets
Seismicprestack PP or PS
modelling
Seismic CMPgathers
FORCE seminar - 06/02/02. Prestack Data. NPD Stavanger.
Philippe JULIEN - Massive modelling and AVO
THE MODELLING TECHNIQUESTHE MODELLING TECHNIQUES
Details on the 3 modelling algorithms
Reflectivity calculation + Zoeppritz: fast and easy.Can not deal with tuning.Can not deal with internal multiples, complex interference patterns, attenuation.
Ray tracing + Zoeppritz:fast and more precise.Can deal with tuning.Can not deal with internal multiples , complex interference patterns, attenuation.
Wave-equation:Slow but more preciseCan deal with tuningCan deal with internal multiples, wave conversion modes, Q attenuation,...
FORCE seminar - 06/02/02. Prestack Data. NPD Stavanger.
Philippe JULIEN - Massive modelling and AVO
OVERVIEW OF THE PRESENTATION
I) A few recalls: Petro-elastic modelling, amplitudecalibration and seismic massive modelling approach.
II) Application: A real data example: impact onacquisition, processing and interpretation.
III) 2D extension: the approach, a real case, impact onthe interpretation.
FORCE seminar - 06/02/02. Prestack Data. NPD Stavanger.
Philippe JULIEN - Massive modelling and AVO
1
2
5
9
13
Ocean water
Background
Transition zone
Transition zone
Transition zone
Sand A
Sand B
Sand C
Surrounding shale
General geometry of the reservoir model
Perturbations:For each sand separately
PHI: 10% -> 30% step 5%Sw: 20% to 100% step 40%Vclay: 5% to 35% step 10%thick: 5m to 40m step 5m
choosing only models such asPHI / (1-Vclay) <=0.4 (petrophy. condition)
TOTAL for each sand: 792 models
TOTAL number of models:2376
AVO ANALYSIS OF POSSIBLE RESERVOIRS
FORCE seminar - 06/02/02. Prestack Data. NPD Stavanger.
Philippe JULIEN - Massive modelling and AVO
Impact of parameters on AVO response of sands
Water saturation increasing
Porosity increasing
« Vclay increasing »
Near
Far
Near
Farthickness
increasing
Water line
SUMMARY OF THE STUDY
A geological input is necessary.
Arrows indicate impact of modfying eachparameter on the AVO near-far response.
Porosity has a huge impact on amplitudes.Sw and Vclay have an impact on AVO.
Modifying (PHI and Sw) can leadto the same seismic responseas modifying Vclay.
FORCE seminar - 06/02/02. Prestack Data. NPD Stavanger.
Philippe JULIEN - Massive modelling and AVO
elastic 1
elas
tic 2
2
PetrophyPetrophy. X. XSeismic att. 1
Seis
mic
att.
2PetrophyPetrophy. X. X
Seismic att. 1
Seis
mic
att.
2
PetrophyPetrophy. X. X
UPSCALING
3rd action: seismic
2nd action: model
1st action: well INCREASE OF UNCERTAINTIES
Main methodology for Seismic Reservoir Characterization
FORCE seminar - 06/02/02. Prestack Data. NPD Stavanger.
Philippe JULIEN - Massive modelling and AVO
IMPORTANCE OF STATISTICAL CALIBRATION
AVO calibration & seismic modelling have an impact on: 1) Seismic acquisition: * maximum offset ==> de-multiple techniques,
* necessary seismic resolution (i.e. freq. content), * acquisition foot-print reduction,…
FORCE seminar - 06/02/02. Prestack Data. NPD Stavanger.
Philippe JULIEN - Massive modelling and AVO
AVO analysis with massive seismic AVO analysis with massive seismic modellingmodellingEXAMPLE OF APPLICATIONEXAMPLE OF APPLICATION Example : Post-Mortem StudyExample : Post-Mortem Study
All models with phi below 20%,No tuning (overburden shale layer and reservoir >150 m)
Gas(Sw20/60%)
Brine
• Fluid discrimination difficult as brine case (phi=10%) is very similar to gas response (phi=15%). May be possible on long offsets.
Impact of fluid contentImpact of fluid content
FORCE seminar - 06/02/02. Prestack Data. NPD Stavanger.
Philippe JULIEN - Massive modelling and AVO
IMPORTANCE OF STATISTICAL CALIBRATION
AVO calibration & modelling have an impact on: 1) Seismic acquisition: * maximum offset ==> de-multiple techniques,
* necessary seismic resolution (i.e. freq. content), * foot-print reduction,…
2) Seismic processing: * choice of the useful offset ranges, * reliability of amplitudes and Signal/Noise ratio
improvement, * full-wave modelling shows how seismic wave
propagation (multiples, converted waves, refractions,…)does [not] disturb amplitude analysis.
FORCE seminar - 06/02/02. Prestack Data. NPD Stavanger.
Philippe JULIEN - Massive modelling and AVO
IMPORTANCE OF STATISTICAL CALIBRATION
AVO calibration & modelling have an impact on: 1) Seismic acquisition: * maximum offset ==> multiple removal,
* necessary seismic resolution (i.e. freq. content), * acquisition foot-print reduction,… 2) Seismic processing: * choice of the useful offset ranges, * reliability of amplitudes and Signal/Noise ratio improvement,… * impact on amplitudes of seismic wave propagation.
3) Seismic interpretation: * ranking of petrophysical parameters, * building a list of possible AVO cases (==> What are
the best seismic attributes?), * quantitative interp. & uncertainties eval. & reduction, * seismic monitoring: impact of pressure &
saturation,...
FORCE seminar - 06/02/02. Prestack Data. NPD Stavanger.
Philippe JULIEN - Massive modelling and AVO
OVERVIEW OF THE PRESENTATION
I) A few recalls: Petro-elastic modelling, AVO calibration,and 1D seismic massive modelling approach.
II) Application: A real data example: impact onacquisition, processing and interpretation.
III) 2D or 3D (1.5 D) perturbations: the approach andexemples: impact on the interpretation.
FORCE seminar - 06/02/02. Prestack Data. NPD Stavanger.
Philippe JULIEN - Massive modelling and AVO
SIMPLE APPLICATION ON A REAL DATA EXAMPLE
Create prestack synthetic images from geological facies interpretation
Two possible fluid scenarii:- water- light oil
FORCE seminar - 06/02/02. Prestack Data. NPD Stavanger.
Philippe JULIEN - Massive modelling and AVO
SIMPLE APPLICATION ON A REAL DATA EXAMPLE - Synthetic versus real : analyse differences and similarities
LIGHT OIL CASE
FORCE seminar - 06/02/02. Prestack Data. NPD Stavanger.
Philippe JULIEN - Massive modelling and AVO
SIMPLE APPLICATION ON A REAL DATA EXAMPLE
Synthetic versus real : analyse differences and similarities
WATER CASE
FORCE seminar - 06/02/02. Prestack Data. NPD Stavanger.
Philippe JULIEN - Massive modelling and AVO
MODELING OF PERTURBED PETROPHYSICAL AND STRUCTURAL SCENARII MODELING OF PERTURBED PETROPHYSICAL AND STRUCTURAL SCENARII Description of the initial model and perturbations from X wells.
Water
Overburden
Layer A
ReservoirLayer BLayer C
Basement
Sw: 25 - 55 % dSW: 10 %Porosity: 18 -30 % dpor: 4%Structure:
✔ Overburden: constant (fromseabottom)
✔ Layer A(shale): 5-20 m delta 5m✔ Reservoir (Gas): 9 m -29 m delta:
5m✔ Layer C.: 0.2 - 50.2 m delta: 25 m
960 perturbations
Application on dataset
Seismic modeling of synthetic CMP gathers...Seismic modeling of synthetic CMP gathers...
FORCE seminar - 06/02/02. Prestack Data. NPD Stavanger.
Philippe JULIEN - Massive modelling and AVO
Classification of real seismic data using post-stack synthetic wavelets.
Gas ThicknessGas Thickness Porosity MapPorosity Map
Seismic modeling of synthetic CMP gathers...Seismic modeling of synthetic CMP gathers...
FORCE seminar - 06/02/02. Prestack Data. NPD Stavanger.
Philippe JULIEN - Massive modelling and AVO
Well 1Well 1
Gas: Gas: measured: 14 m measured: 14 m predicted: 13.2 m predicted: 13.2 m
Porosity: Porosity: measured: 24%-29% measured: 24%-29% predicted: 29.5% predicted: 29.5%
Well 2Well 2
Gas: Gas: measured: 14 m measured: 14 m predicted: 14.9 m predicted: 14.9 m
Porosity: Porosity: measured: 23%-28% measured: 23%-28% predicted: 26.1 % predicted: 26.1 %
Well 3Well 3
Gas: Gas: measured: 2 m measured: 2 m predicted: 6.6 m predicted: 6.6 m
Porosity: Porosity: measured: 28%-30% measured: 28%-30% predicted: 26.0 % predicted: 26.0 %
PREDICTION OF RESERVOIR PROPERTIES AT WELL LOCATIONSPREDICTION OF RESERVOIR PROPERTIES AT WELL LOCATIONS
Seismic modeling of synthetic CMP gathers...Seismic modeling of synthetic CMP gathers...
FORCE seminar - 06/02/02. Prestack Data. NPD Stavanger.
Philippe JULIEN - Massive modelling and AVO
CONCLUSION
1) For many years, structural calibration has beenmainly highlighted. AVO calibration is a mandatoryprocess.
2) The AVO phenomenom is very complex: the solutionis generally not unique.AVO calibration (including petro-elastic modelling &massive seismic modelling) allows to understand whyso many dry wells have been drilled, using AVO.
FORCE seminar - 06/02/02. Prestack Data. NPD Stavanger.
Philippe JULIEN - Massive modelling and AVO
CONCLUSION
3) We propose a methodology which goes beyond theconventional calibration:
- to identify the seismic response of a geologicalhypothesis: to be more predictive,
- to test different scenarii ==> uncertainties reduction, - to look at the impact on acquisition, processing and
interpretation.
4) We already knew that seismic calibration was onemeeting point between geologists & geophysicists.
Amplitudes calibration (broadly speaking) is ameeting point in geosciences