A compressional wave seismic model for the Ashrafi-1 well

Abstract Despite disappointing drilling results, the South Caspian basin has yielded scientific results of significant value to future exploration. This paper describes work that utilized data from the offshore Ashrafi-1 well, but we believe it has applications to exploration elsewhere in Azerbaijan. The North Absheron Operating Company (NAOC) drilled the Ashrafi-1 well between November 30, 1997 and February 6, 1998 to a total depth of 3668 m (sub Caspian sea level). The Balakhany and Pereryva Suite (SP) Formations were wet but the Upper Kirmaku Sandy (NKP) Formation tested oil and the Lower Kirmaku (PK) Formation tested gas, both at high flow rates. Anomalously high stack amplitude responses, generally conforming to structural closure, had been observed in the 3D seismic data at the NKP and PK Formations prior to drilling of the well. An offset synthetic modeling study was undertaken, post-drill, to determine the in-situ amplitude response (stack and amplitude variation with offset, or AVO) for both pay and wet zones and to model the changes in AVO and stacked amplitudes in response to changes in variables such as thickness, fluid fill and porosity. As well as the 3D seismic data set, the principal data sources used were a dipole sonic log for compressional wave velocity ( V p) and shear wave velocity ( V s); checkshots for time-to depth ( T – D ) conversion; a density log; log interpretations of porosity; fluid properties based on pressure–volume–temperature (PVT) analysis of fluids from drill stem tests (DSTs) and pressure/temperature data. In this simple case, despite the highly permeable reservoir rocks, the Gassmann assumption of immobile fluids within the pore space was used. Although the effects of partial gas saturation are known to be a problem elsewhere in this basin, density modeling derived from borehole-guided long offset AVO was not attempted. Despite this, normal incident amplitudes, derived from pre-stack data, cross-plotted against amplitude gradients suggest a methodology for distinguishing wet sandstones from gas and oil in this basin in the absence of recent leakage. Despite the difference in response for oil and gas (oil gave a stronger response than gas), a statistically meaningful number of wells is not yet available and this result may be influenced by local lithologic effects. Thus, the ability to distinguish oil from gas ahead of the drill bit — which is of vital importance in evaluating commerciality — is not proven.