다중 탄성파 속성 분석을 이용한 시추공이 없는 지역에서의 최적 시추위치 선정

Abstract

For Quantitative Seismic Interpretation (QSI) of seismic data, it is important to obtain well log data which provides elastic properties of the subsurface formations. However, in many fields, only seismic data are available and the cost of drilling is enormous. Therefore, in this study, the workflow which estimates the distribution of hydrocarbon-bearing sand bodies in the region having no well log data was suggested using multi-attribute analysis and sparse-spike inversion with RMS velocity information. The methodology and results of suggested workflow were presented based on the applications of the workflow to two field data: Case 1 – gas hydrate reservoirs in Ulleung Basin, East Sea and Case 2 – deep-water conventional oil and gas reservoirs with minor volcanism. The workflow consists of two phases: Phase 1 - identification of candidate zones using multi-attribute analysis and Phase 2 - determination of sweet spot using impedance inversion for estimating the extent and thickness of potential reservoirs. In the first phase, seismic attributes which show unique characteristics of gas hydrate and hydrocarbon reservoirs more than typical indicators such as BSR and bright spot were selected, considering the elastic properties of target resources and surrounding geology. After finding candidate zones in the field through the attributes, two steps of impedance inversion were performed: Band-limited inversion and Sparse-spike inversion. Firstly, wavelets were extracted with statistical extraction method and low frequency model was built from RMS velocity information using Dix equation and Gardner’s equation. Secondly, acoustic impedance model was inverted by the integration of low frequency components from RMS velocity and seismic frequency components from input traces using band-limited inversion based on recursive algorithm. Thirdly, sparse-spike inversion was conducted to improve the model by eliminating the meaningless reflectivity series from side-lobes of wavelets and random noise. As a result, the improved model had better layering for visualization and lower RMS error than previous inverted result. This model was also verified through the comparison of the acoustic impedance log and inverted model from conventional model-based inversion with high correlations. In the Case 1, sweet spot was identified with the area of higher RMS amplitude, higher instantaneous frequency, lower structural discontinuities, and lower acoustic impedance upon BSR. In the Case 2, the promising well site was identified with the clastic reservoirs with oil and associated gas in fault zones which show higher RMS amplitude and lower acoustic impedance. Consequently, the sweet spots for two cases were successfully determined, and the optimized workflow was proved to be plausible for identification of sweet spot as well as the distribution of reservoirs.