Overpressure detection using shear-wave velocity data: a case study from the Kimmeridge Clay Formation, UK Central North Sea A. Edwards, S. O Connor, S. Green, and K. Waters, Ikon Science UK
Overview Introduction The problem Secondary processes Shear-wave data Case study : Kimmeridge Clay Formation, UK Central North Sea Study database Identifying mechanisms Vp vs. effective stress trends Vs vs. effective stress trends Vs-based pore pressure predictions Conclusions and implications
The problem The slowing of compressional-wave velocity (Vp) with increasing pore pressure has been the basis for overpressure detection and quantification in shales for many years Typically, an increase in pore pressure (more accurately recognised as an increase in overpressure; the difference between the pore pressure and the hydrostatic (normal) pressure) as a result in ineffective dewatering/fluid escape arrests compaction, leading to anomalously high porosity for a given depth of burial The success of pore pressure prediction, and consequently overpressure quantification, is highly dependent on the mineralogy, depth of burial, and the diagenetic history of the shales, that is, the shale velocities can be non-unique with respect to the pore pressure Overpressure quantification is complicated in areas where disequilibrium compaction is not the active, or not the only active, process such as: In hot shales, e.g. High Pressure and High Temperature (HP/HT) plays such as, the Jurassic source rock in the Central North Sea Resource plays such as, the gas-bearing shales in the Bossier and Haynesville formations in Northwest Louisiana
Bowers velocity / effective stress model (after Bowers, 1995; Tingay et al., 2008)
Identifying secondary processes (after Hoesni, 2002; Swarbrick, 2012)
Effect of TOC on wireline response Log Type TOC 0% TOC 3% TOC 10% Rho (g/cm3) Sonic (us/ft) Vp (m/s) 2.39 2.32 2.14 102 116 139 2988 2627 2192 Rock Physics can be used to perform; a) Fluid substitution (replace gas with brine and simulate the wireline response b) Produce a pseudo-vp based on the Vs as this is unaffected by gas 10% TOC The reduction in Density can be 250 kg/m 3 The reduction in Velocity can be 740 m/s (after Passey et al., 1990)
Gas-prone rocks: challenges Compressional velocity data can be significantly reduced by the gas, hence, it may not be suitable for meaningful pressure predictions in shales without correction Alternative methods of pressure prediction involve: Use density (not gas affected) to estimate pore pressure? Use density (not gas affected) to estimate sonic via Gardner transform? Use Vs data (not gas affected) to estimate Vp using relationships from unaffected analogues? Use Vs data directly by generating a locally calibrated Eaton relationship? Use a velocity-effective stress relationship?
Ebrom et al. (2003) Advantages of Vs data: Not significantly affected by fluid presence, e.g. free gas, as the shear velocity magnitude reduces slightly but the shear modulus (G) remains constant More sensitive to low effective stress (or high pore pressures) (Dutta, 2002; Bakulin et al., 2008) Extended traditional methods to non-traditional data sources, i.e. Vs to escape the effects of gas Offshore Trinidad Conventional setting but well was drilled through a gas chimney
Zhang and Wieseneck (2011) Haynesville & Bossier Plays Extended traditional methods to non-traditional data sources, i.e. built a Vs to Vp transform to remove the gas effects
Couzens-Schultz et al. (2013) Large scatter in Vp values Haynesville & Bossier Plays Extended traditional methods to non-traditional data sources, i.e. Vs Produced accurate predictive models
Case study: Kimmeridge Clay Formation, UK Central North Sea
Study database The well database consisted of 35 wells covering UK Central North Sea Quadrants 21, 22, 23, 28, 29, 30, 31, 38 and 39 All well data were collated into a single definitive database All Wireline Formation Test data were interpreted from original build-up plots, giving accurate formation / pore pressure in porous units These resulting data were used to derive fluid gradients where appropriate to determine reservoir overpressure Other less reliable indicators of formation pressure were connection gas and kicks (after Ikon Science, 2014)
Origin and distribution of Jurassic and Triassic overpressures The origin of the overpressure is considered to be a function of rapid loading during late Cenozoic burial as well as contributions from thermal processes (e.g. gas generation) during deep, high temperature burial (Cayley, 1987; Holm, 1995; Swarbrick et al., 2002). The overpressure is distributed mainly in Jurassic and Triassic pressure cells with the magnitude increasing with depth. The highest overpressures (>8000 psi) are found in the East Central Graben. An exception to the pattern of increasing overpressure with increasing depth of burial is the area in which the Fulmar, Halley and Clyde fields are located (UK Blocks 30/11, 30/12, 30/16 and 30/17). The lowoverpressure area has anomalously low Jurassic Shoreface reservoir pressures (<1000 psi) relative to depth of burial, and by comparison with sediments above The challenge: is to predict the Jurassic reservoir pressures as accurately as possible in the Kimmeridge Clay Formation (high temp, high TOC, gas saturated)
Vs (m/s) Vs (m/s) Tertiary - Jurassic Vp / Vs trends 3500 1000 1500 3500 Jurassic shales show anomalously low Vp/Vs at increasingly high temperatures 2000 2500 3000 2000 3500 2500 4000 4500 3000 5000 3500 5500 500 500 1500 6000 1500 6000 Vp (m/s) Tertiary shales show a relatively uniform Vp/Vs behaviour which trends with temperature Depth (m) Therefore, above a certain temperature, reservoirs may no longer show the typical low Poissons ratio response! Vp (m/s) Depth (m) Jurassic shales, show a strong variation in Vp/Vs behaviour here we see good correlation to temperature. This is suggestive of a hydrocarbon effect the stepping nature being related to the transition from immature gas generating!
Vp / vertical effective stress trends The data show a large scatter in the Vp values likely indicating that individual velocity/stress relationships are needed per well rather than a two regional Vp trends for shallow and deeper data Regional Vpunloading trend Regional Vploading trend Circles indicate individually picked thick (>10 m) shale packages
Vs / vertical effective stress trends To link Vs logs to pore pressure we need to establish a velocity (from log) vs. effective stress model. To this end, two approaches were taken to generating shear log (Vs) vs. VES: 1. Using direct pressure data, i.e. kicks and wireline formation test (WFT) data measured in Jurassic-age sands in the Kimmeridge Clay that are considered to represent un-drained virgin pressures, to generate a velocity/stress trend 2. In the second approach, the pore pressure input required to give VES has been taken from mud-weights rather than WFT data used to drill the selected wells. Only wells that are close to balance that is, wells that have evidence of connection gases etc. have been used. In these selected wells, the static mud-weight plotted is considered equivalent to formation pore pressure or very close to it.
Vs / vertical effective stress trends Circles indicate individually picked thick (>10 m) shale packages Regional Vs-trend A single velocity/stress trend is produced to match Jurassic data from all wells In reality there may be, and should be, a series of trends as each shale in each well will unload differently; however, this study attempts to produce a regional model which subsequently is applied to a series of blind test wells If successful, the model derived here then becomes a predictive tool at a prospect location. If locally calibrated solutions were built per well then there would be no predictive capability for prospect locations
Vs-based pore pressure prediction
Vs-based pore pressure prediction Regional Vploading trend Regional Vs-trend Regional Vpunloading trend
Vs-based pore pressure prediction Corroborate the Vs results with regional aquifer overpressure maps (after Ikon Science, 2010) 2014)
Vs-based pore pressure prediction Regional Vploading trend Regional Vs-trend Regional Vpunloading trend
Testing the Model Regional Vploading trend Regional Vs-trend Regional Vpunloading trend
Testing the Model Regional Vploading trend Regional Vs-trend Regional Vpunloading trend
Testing the Model Regional Vploading trend Regional Vs-trend Regional Vpunloading trend
Conclusions and implications We have presented a model to predict pore pressure in the Kimmeridge Clay Formation in the Central North Sea Using Vs logs mitigates against the gas effect; furthermore S-waves are more sensitive to changes in effective stress, particularly where the stress field is isotropic (as in HP/HT conditions) thus resulting in a more accurate pressure interpretation. The application of the technique described in this paper: Can save time and cost conducting a fluid substitution for multiple wells Describes a scientifically sound workflow Can overcome the difficulties in predicting shale pore pressures in HPHT environments, source shales and areas where overburden is not the maximum vertical stress Provides a simpler workflow for wider use within the industry One limitation of the method evolves around how a Vs log / volume is derived particularly pertinent in data limited areas (e.g. frontier basins)
Dr. Alexander Edwards aedwards@ikonscience.com +44 (0) 20 111 222 333