Devex’09 13 May 2009 1

Correlations in well rate fluctuations and coupled geomechanical modelling used to identify reservoir communication paths

John Greenhough, Ian Main, Stuart Simmons The University of Edinburgh

Kes Heffer, Reservoir Dynamics Ltd

Xing Zhang, Assef Mohamad Hussein, Gaisoni Nasreldin, Nick Koutsabeloulis, Schlumberger Reservoir GeoMechanics Center of Excellence

Devex 092

Talk outline

• The Statistical Reservoir Model– Short­term forecasting

• Rate correlations – General characteristics– Conceptual mechanisms

• Rate diffusivities• Valhall results (cf. microseismic records)• Aggregated results from 6 North Sea fields• Integration into reservoir management• Messages

Devex 093

Flow rate fluctuations

Producers

Injectors

Devex 09

Statistical Reservoir Model

• Finds best small group of wells to model flow rate of any well of interest

Devex 09

Example of correlated wells

Devex 09

History Match Well 15 (R2!"#$%&'!()*

Devex 09

Statistics of fits & forecasts

Total production errors ~2 x size if only use previous month as fit

211416117Error ! on total production [%]

2224173317Mean error ! over all producers [%]

0.970.950.960.900.92Mean R2 over all producers

4/26(15%)

9/82(11%)

8/72(11%)

13/59(22%)

12/159(8%)

Mean predictors per producer (out of total predictors with 10+ months data)

16/1770/8137/4834/3673/118Producers output (out of those with 10+ months data)

Field 5Field 4Field 3Field 2Field 1

Devex 098

Short­term rate prediction

Blind test of prediction with SRM of 3 months of Gullfaks production (refereed by Rock Deformation Research, University of Leeds)“The actual production figures for most (>70%) of the wellbores lie

within the 95% confidence limits given by Edinburgh.”

Devex 099

broadband fluctuationshigh frequency fluctuations__ zero correlation

Shmax

First principal component of matrix of rate correlations between all wells in field B – independent mode ‘explaining’ largest proportion of fluctuation variance

General characteristics of rate correlations (Spearman rank)

Long­range

Fault­related

Stress­related

Devex 0910

Basic concept

Devex 0911

Concept of long­range transmission of strain by shear

dilatation

compression

More complexdiscontinuousshear

SH

local cause

of rate changes

transmission

Conductivity determined by normal stress during shear vsrock strength, ductility, damage zone, displacement etc

Long­rangeStress­related

Fault­related

Devex 0912

Extraction of rate diffusivities from time­behaviour of rate correlations

Looking to extract tensors related to diffusive rate behaviour:

–Local (shorter­range)

–Non­oscillatory in time

–Fitted with non­negative diffusivities

– (Decaying in time)

Likely to be related to permeability tensors.

Devex 0913

Valhall fielddiffusivity major axes & lag­0 PC1on fault traces

diffusivity major axes on induced inelastic shear strains (VISAGETM)

Devex 0914

Valhall field: Microseismic emissions vs rate diffusivity trends

Devex 0915

Valhall: Microseismic events and interpreted diffusivity axes from rate correlations

Fast shear waves = fracture strikesTeanby et al. (GJI 2004)

Zoback & Zincke(PAGeophys, 2002) Focal mechanism: normal slip

Devex 0916

Valhall: Microseismicity and modelled strain due to depletion

(VISAGE coupled model)

Devex 0917

Valhall: Diffusivity axes and casing collapse/deformation

Devex 0918

Neotectonic setting of North SeaMaximum horizontal stress axes(World Stress Map Heidbach et al (2008))

Earthquake epicentres (US Geological Survey/National Earthquake Information Center (PDE) 1973 ­ Present)

Devex 0919

Aggregated North Sea fields­ Long­range correlations (zero lag time)(detrended rate data)

From Statistical Reservoir Model:

Mean exponent of ratio correlated/available wellpairs vs distance = ­0.14

Exponent ~ ­0.07

Average exponent of mean correlated distance vs time ~ 0.1

cf Fickian diffusion = 0.5 => NON­FICKIAN DIFFUSION

Devex 0920

Boundary effect?

Coupled VISAGE model of Valhall:

Normal shear strain in top reservoir layer

Raw rate data

Devex 0921

Orientational distribution of major diffusivity axes (aggregated fields)

Shmax

North Viking graben South Viking / Witch Ground grabens

Central graben

N SStrike­slip/thrust stress state

Highly anisotropic horizontal stresses

Normal stress state

~Isotropic horizontal stress

Azimuthally smoothed Slip tendency of faults

Devex 0922

Orientational frequency distributions for 2 types of analysis of production histories both indicate the involvement of shear slip – even in ‘unfractured’ fields

______

most favoured breakthrough directions for injected fluid in 47 ‘unfractured’fields worldwide

________

major diffusivity axes for 4 ‘unfractured’fields in North Sea

Devex 0923

Integration into reservoir management work process

Rate CorrelationsU Edinburgh SRM

Correlation matrix analysisPCA, FA, POP, diffusivities

Interpolate factors across field

Compare with structural knowledge

Reservoir model VISAGE / VIP/ ECLIPSE etc

Validate / calibrate Geomechanical model

Predict / Optimise long­term production

Predict / Optimise short­term production

coupling

Infill well numbers & configuration; pressure management; improved vertical & areal sweep efficiencies; water production reduction; EOR schemes…

Choke settings

Devex 0924

Messages

• Geomechanical effects appear to be influencing fluid behaviour in reservoirs worldwide (these effects appear to be commercially significant)

• Rate fluctuation analysis and geomechanical modelling can assist understand those effects and provide commercially useful information >>>

Devex 0925

Potential application of rate correlation analysis

1. Indicator of reservoir communications, complementary to geological model,microseismicity, seismic anisotropy, 4D seismic, well test permeabilities (mutual validation & calibration).

2. Time­lapse changes in communications .3. Calibration of geomechanical understanding

/model4. Short­term rate predictions5. Rate optimisation

Low cost; no acquisition

Devex 0926

Acknowledgements

NERC CONNECT grant GR3/C0022 with matching funding from BP

COFFERS project, under the Industry Technology Facilitator (sponsors: Amerada Hess, BG Group, BP, Conoco­Phillips, DTI, Kerr­McGee, StatoilHydro, Shell and Total)

RESURGE project, sponsored by theTechnology Strategy BoardFIELD DATA provided by BP, Nexen, StatoilHydro