An Integrated Approach to Delineate Reservoir Facies through Multi Attribute Analysis in Complex Lithological Environment

Abstract

Abstract Multi attribute analysis is a technique through which we predict reservoir parameters for identifying facies. This analysis applied to seismic volume helps not only in identification of reservoir distribution pattern but also gives a fair indication of the quality of reservoir which can be expected in yet to be explored/developed areas. In the study area, low impedance contrast between sand-shale, which is inter-bedded with coal, hinders proper imaging of reservoir geometry. Area of study pertains to Cambay Basin, India where the mapping of pay sands in the areas with less/no well control were done through application of seismic inversion, neutron porosity modeling and final results were corroborated with spectrally decomposed volumes through generation of amplitude maps in zone of interest. Spectral decomposition provides a novel means of utilizing seismic data for imaging and mapping temporal bed thickness and geologic discontinuities in connection with reservoir properties and seismic attributes over large area. A multi mineral log processing approach was adopted for estimation of petrophysical parameters, mineral volumes and fluid contents. A feasibility study has been carried out to see the possibilities of predicting petrophysical parameters from seismic attributes. Cross-plots between P-impedance, neutron porosity (NPHI), Vp/Vs and other logs have been attempted to identify the different lithologies with fluid type. Since the reservoir is very much heterogeneous in terms of variation in lithology and mineralogy, histogram of various log response along with facies and fluids have been also been taken to discriminate the reservoir facies. Low frequency P-impedance model based inversion was carried out after selecting most suited inversion parameters, which help to delineate coal facies in zone of interest. The cross plots between P-impedance and NPHI shows a clear distinction of reservoir and non-reservoir facies. Therefore modeling of NHPI has been attempted through application of Probabilistic Neural Network (PNN). In this technique, various seismic attributes viz. integrated absolute amplitude, second derivative of instantaneous amplitude, quadrature trace, average frequency, dominant frequency, instantaneous frequency, frequency filter, cosine of instantaneous phase and amplitude weighted cosine phase including inversion output (P-impedance volume) has been used to delineate reservoir in optimal manner. The set of these attributes have been trained to predict NPHI at well location. Network thus generated was then applied on the whole seismic volume for all different attributes to predict NPHI volume though PNN. Present study has brought out two areas with good sand development. This methodology presented way of unique combination of reservoir properties and seismic derived attributes through multi attribute analysis that can be utilized where the heterogeneity of fluid content is present. When there is limited number of wells in the area, this methodology is significantly useful for generation of locales/leads for exploration & exploitation strategy in development stages.