Fractures and fluid flow in petroleum reservoirs Quentin Fisher Centre for Integrated Petroleum Engineering and Geoscience School of Earth and Environment University of Leeds E-mail: quentin@rdr.leeds.ac.uk cipeg.leeds.ac.uk
Outline Drilling wells in petroleum reservoirs can cost >$200 million The success of those wells is often dependent on predicting how faults and fractures are affecting fluid flow in the subsurfac Key issues: - Faults: conduits vs barriers Fractures: predicting their distribution
Faults as barriers to flow (from van der Molen et al., 2003 EAGE conference on seals, Montpellier)
Fault rocks as flow barriers
Faults as conduits for fluid flow
Faults as conduits for fluid flow Mud losses
Conduits vs barriers Active vs inactive Often argued that active (or critically stressed) faults are conduits whereas inactive (or below critical stress) are barriers Many examples where seismicity is not associated with significant fluid flow Most published examples where critically stressed faults increase flow are from hard rocks Rheology vs stress conditions Barriers when formed when deformation is ductile Conduits often form when deformation in brittle C.f. critical state soil mechanics
Faults in clean Brent sands Time of deformation Devonian breccias & fractures Brent Faults Rotliegendes cemented faults Rotliegendes Cataclasties & fractures Columbian faults Start of quartz cementation o (~90 C) Depth
Rotliegendes: : cataclastic faults Time of deformation Devonian breccias & fractures Brent Faults Rotliegendes cemented faults & fractures Rotliegendes Cataclasties Depth Columbian faults Start of quartz cementation o (~90 C)
Faults in Colombian sandstone Faulting of very low porosity rocks (<15% porosity) often from conduits not barriers to flow Time of deformation Devonian breccias & fractures Brent Faults Rotliegendes cemented faults & fractures Rotliegendes Cataclasties Depth Columbian faults Start of quartz cementation o (~90 C)
Fault breccia Devonian Reservoirs North Sea Time of deformation Devonian breccias & fractures Brent Faults Rotliegendes cemented faults Rotliegendes Cataclasties Depth & fractures Columbian faults Start of quartz cementation o (~90 C)
Faults as conduits vs barriers in clean sandstones: rule-of-thumb Consistent with observation, critical state theory of soil mechanics, experimental rock mechanics and numerical modelling.
Key uncertainties Impact of fault rocks on fluid flow over geological timescale Controls on the brittle-ductile transition in shales Lack of experimental data Calibration of fault seal prediction methodologies Most ways of calibrating fault seal predictions are intrinsically non-unique Complexity Fault zones tend to be very complex but we model them in a very simple way is this OK?
Fracture prediction Mode I fractures often act as conduits for fluid flow in petroleum reservoirs Often flow is highly localised In fractured reservoirs, a huge proportion of oil production often comes from very thin intervals in a small proportion of wells Predicting the presence/distribution of fractures is intrinsically difficult Fracture models often don t work but could we do better? Are seismic methods best way of predicting fracture distribution? How do we predict the long term (>50 years) changes in fracture permeability as a function of change in reservoir pressure (tight gas sands)
Top seal capacity Filled to spill Under-filled Empty
Capillary seals 2R Where: P c Pc = 2γcosθ/R 2σ cosθ = R P th = threshold pressure (psi) σ = interfacial tension (Dynes/cm) θ= contact angle R = pore throat radius (microns) Top seals have small pore-throat sizes and therefore can act as capillary seals
Leakage along hydrofractures Pore pressure needs to overcome minimum horizontal stress while leakage occurs From Nordgård Bolås and Hermunrud, 2003
Problems with existing methodologies The petroleum industry is not very good at predicting top seal failure OK at post-mortems Do we actually know how top seals leak? Can we predict top seal leakage during reinjection of CO 2? Could seismic help? What modelling should be used?
Stress path during re-inflation Effective stress σ = S α pδ Stress path parameter γ 3 ij ΔS3 = Δp ij ij σ ij effective stress S ij total stress α Biot coefficient p pore pressure Estimates of stress path have been made from repeated leakoff tests during depletion Some evidence that stress paths can be lower during inflation than deflation (i.e. fracture pressure is lower) From Santarelli et al., (SPE, 47350)
Modelling fracture etc MPI interface Synthetic seismic Coupled geomechanical production simulation models
Valhall Field - Background (from Barkved, 2003) (from Kristiansen, 1998)
Seismic anisotropy & shear wave splitting Seismic anisotropy is the directional dependence in seismic velocities - Indicator of order in a medium - Indicator of style of flow, stress regime or fracturing Shear-wave splitting
Fracture size estimation using frequency- dependent shear-wave splitting. After Maultzsch et al. (2003); EAP work
Yibal: Results for carbonate reservoir. Clear freq-dependent anisotropy Yibal Valhall: Results for overburden Low amount of anisotropy No obvious freqdependent anisotropy Valhall
Conclusions We seem to be making significant progress predicting the fluid flow properties of faults in petroleum reservoirs Calibration is difficult due to non-uniqueness of evidence More work is needed to understand how faults affect fluid flow in sediments such as shales Predicting the distribution of open fractures in the subsurface is more challenging Fracture models often don t work Maybe seismic methods are more promising Could we improve our modelling?