Detecting a gas injection front using a 3D seismic data: case of an Asmari carbonate reservoir in the Zagros basin

Abstract

Extracting reservoir properties from a 3D seismic cube is an extremely sensitive job. While the inversion procedure for obtaining the inverted seismic data is itself an important step, most of such sensitivity contributes with applying multi-attribute transforms on an inverted cube to obtain a cube of reservoir property. A multi-attribute transform was applied on a 3D seismic data of a gentle anticline located southwest of Iran. Few well logs from the full sets were missing which were estimated using neural networks with sufficiently large correlations. An extended study was performed to extract the most suitable wavelet from the seismic data. The 3D cube was inverted using the model-based inversion method. The inverted cube along with other extracted cubes of seismic attributes was used as inputs to neural networks to estimate reservoir properties. Various attribute lists and estimation methods were applied and studied, and the best set of attributes and the best method of estimation were introduced. Porosity distribution was responsible for pressure difference across the reservoir; pressure communication between two upper zones was approved; results demonstrated a new potential hydrocarbon-bearing layer separated from the above producing layer by a nonporous layer. The existence of such a potential was approved by drilling a new well. The extent of reservoir flooded by the injected gas was demonstrated.