Time-Lapse Seismic: A Geophysicist s Perspective Rob Staples Shell E & P Europe SPE / EAGE workshop Copenhagen, March 2004
1. What do we observe? 2. Noise and detectability 3. Resolution, scale, interference 4. Other problems 5. Working domain and communicating results
Saturation changes Acoustic impedance oil water gas Depth 90 phase rotation + Possible spectral adjustment Reflectivity difference data Loop approx. at top and base of flood zone Pseudo-Impedance difference data Loop approx. represents flood zone
Saturation changes Acoustic impedance oil water gas Depth 90 phase rotation + Possible spectral adjustment Reflectivity Pseudo-Impedance difference difference data data Loop Loop at top approx. and base represents of flood zone flood zone Body-checking Pseudo-Impedance of difference cube data Quick & approximate Loop represents voxel mapping flood zone of flood
Saturation changes Amplitude of saturation changes depends on: Initial / residual saturation Oil properties or gas properties Porosity Net-to-gross Rock frame stiffness
Pressure-up (injection) Normalised Velocity 1.1 1 0.9 0.8 0.7 0.6 Initial state Core measurements 0.5 0.4 Increasing pore pressure This zone softens due to injection Axial Stress MPa pressure Seismic section through TA27S2 0 5 10 15 20 25 30 35 40 Observed seismic Upper Ness perforated injection interval 1000ft Lower ness perfs Note: changes generally limited to near-reservoir Rannoch perforations
Relaxation after injection Softening = pressure up 2002 1998 1.1 Normalised Velocity 1 0.9 0.8 0.7 0.6 Initial state 0.5 0.4 Decreasing pore pressure 0 5 10 15 20 25 30 35 40 Axial Stress MPa Core measurements After pressure-up prior to 1998, this area relaxes between 1998 and 2002 causing a hardening Hardening = relaxation
Depletion (below initial pore pressure) Total time shift Shale dt +ve Reservoir dt -ve Reduced PP => increased eff stress => small velocity increase + very small compaction Shale dt +ve Underburden and overburden moves in to fill space. If limited lateral extent, causes tensile stress to shales => small velocity decrease over thousands of feet => time shifts z
Depletion (below initial pore pressure) Data taken from Paul Hatchell et al, SEG2003, paper RCT1.1, (73rd SEG meeting, p1330-1333). Time shifts (blue= pushdown, red=pull-up) extend thousands of feet above and below reservoir Geomechanical prediction 1 km 1 Sec. 1 Sec. 1 km Time-lapse seismic
1. What do we observe? 2. Noise and detectability 3. Resolution, scale, interference 4. Other problems 5. Working domain and communicating results
Sources of 4D noise 1. Marine acquisition changes sampling different overburden 1. Different shot locations 2. Differential streamer feather 3. Undershooting platforms / vessels 4. Different cross-line geometry 5. Differing streamer / source depths 6. Tides => different multiples 2. Land acquisition changes sampling different overburden 1. Different geometry 2. Different water tables 3. Different shot/receiver coupling 3. Differing processing flows 4. Environmental noise
Measuring repeatability RMS (non)-repeatability ratio or normalised RMS Repeatability = RMS difference cube = RMS (monitor base). RMS single survey ½ (RMS mon + RMS base) Measured away from area of true 4D change Typical numbers: Acquisition shot in different directions: 0.6 1.2 (or 60% - 120%) Parallel surveys, legacy processing: 0.4 1.0 Dedicated survey & processing: 0.1 0.4 A lower number is a better survey.
What can we detect? Detection limits depend both on the survey repeatability and on the strength of the 3D earth reflectivity sequence. However, for a typical Northern North Sea geology and acquisition / processing giving repeatability of 20 30%. 0 2 4 6 8 AI change (%) 7% AI change (0.035 reflectivity) clearly visible 5% AI change (0.025 reflectivity) should be detected 3% AI change (0.015 reflectivity) difficult to detect 0 1 2 Time shift (ms) a 2 ms time-shift will be clearly visible a 1 ms time-shift detectable
What can we detect? Detection limits depend both on the survey repeatability and on the strength of the 3D earth reflectivity sequence. However, for a typical Northern North Sea geology and acquisition / processing giving repeatability of 20 30%. Best non-permanent marine measurements 0.3 ms, Maui, NZ 0 1 2 Time shift (ms) Examples shown are not a lower limit. Projects with 10% repeatability and better behaved overburden naturally allow detection of even more subtle effects. a 2 ms time-shift will be clearly visible a 1 ms time-shift detectable
1. What do we observe? 2. Noise and detectability 3. Resolution, scale, interference 4. Other problems 5. Working domain and communicating results
Vertical resolution Scenarios which may be hard to distinguish on typical quality 4D seismic especially at depth 30 120 ft 30ft of flood with 20% residual oil saturation 3 flooded zones 10ft thick of 20% residual oil saturation 50ft of flood with 40% residual oil saturation
Tuning response Actual thickness of modelled wedge of water flooding Apparent thickness of wedge on seismic Note: amplitude increases with thickness (up to a point)
Interfering layers complicate tuning issue Brown only flooding Brown flooded, Blue flooding Blue flooded, Brown flooding Blue only flooding Brown unit Blue unit No flooding Brown flooded Both units Blue flooded No flooding flooded
Interfering layers Brown flooding Amplitude map of measurements made at Brown level. Amplitudes may be strongly affected by Blue flooding (destructively or constructively depending on spacing). Model based stochastic inversion particularly good tool for solving such problems, by calculating range of sweep scenarios that can fit the data. Blue flooding