This paper presents a novel workflow for seismic net pay estimation with uncertainty. It is demonstrated on the Cassra/Iris Field. The theory for the stochastic wavelet derivation (which estimates the seismic noise level along with the wavelet, time-to-depth mapping, and their uncertainties), the stochastic sparse spike inversion, and the net pay estimation (using secant areas) along with its uncertainty; will be outlined. This includes benchmarking of this methodology on a synthetic model. A critical part of this process is the calibration of the secant areas. This is done in a two step process. First, a preliminary calibration is done with the stochastic reflection response modeling using rock physics relationships derived from the well logs. Second, a refinement is made to the calibration to account for the encountered net pay at the wells. Finally, a variogram structure is estimated from the extracted secant area map, then used to build in the lateral correlation to the ensemble of net pay maps while matching the well results to within the nugget of the variogram. These net pay maps are then integrated, over the area of full saturation gas, to give the GIIP distribution (Gaussian distributions for the porosity, gas expansion factor, and gas saturation for the sand end member are assumed and incorporated in the estimate of GIIP). The method is demonstrated on the Iris (UP5 turbidite) interval. The net pay is corrected for reduction in the amplitudes over part of the area due to shallow gas. The sensitivity of the GIIP to the independent stochastic variables is estimated (determining the value of information) so that business decisions can be made that maximize the value of the field.

No abstract

No abstract

No abstract

The Thrace Basin that is located in northwestern Turkey contains sandstone and carbonate reservoirs of Eocene and Oligocene age. Production and exploration activities are still underway. Mapping undrained sweet spots from seismic data is currently a challenge, so time lapse (4D) seismic is used to reduce the risk for new production and development drilling. We have evaluated the normalization and amplitude variation with offset (AVO) analysis of 3D-4D land seismic data in a gas producing field from which baseline and monitor surveys were acquired in 2002 and 2011, respectively. Through AVO analysis, intercept (A) and gradient (B) analysis was conducted, and fluid factor (FF) attribute maps were generated for the assessment of the remaining potential areas. Synthetic gathers were created for simulation of the AVO response, drained and undrained stages and compared with the corresponding 4D seismic data. The drainage of gas from the reservoir interval is evident from the difference maps between 2002 and 2011 seismic data. Both data sets were processed using an amplitude friendly processing sequence. This parallel processing was followed a mild data conditioning and crossequalization for reliable 4D interpretation. The 4D seismic data, especially land data, has low repeatability and requires conditioning to reduce the 4D noise. The 4D noise can be described as nonrepeatable noise, and any difference outside the reservoir zone is not related to production. A so-called crossequalization was applied to the base and the monitor data to bring out similarities so that they cancel out when differences of seismic data and its attributes indicated only the production results over the reservoir zones. As the available 4D data crossequalization software was implemented for stack data only, we created angle band stacks and crossequalized each angle band stack from the base and the monitor data cubes. Five angle band stacks from the base and the monitor prestack data cubes 0°–55° (0°–15°, 15°–25°, 25°–35°, 35°–45°, and 45°–55°) were crossequalized individually. The crossequalized angle band stacks were used in AVO analysis and AVO inversion to generate pore fill identifiers such as FF to map possible undrained zones after 10 years of production.