Calibration of seismic and well data: Towards Improved Quantitative Seismic Reservoir Characterisation of the Triassic to Middle-Jurassic Gullfaks Reservoir Units of the northern North Sea

Abstract: Characterization and evaluation of (oil and gas) reservoirs is typically achieved using a combination of seismic and well data. It is therefore critical that the two data types are well calibrated to correct and account for the fact that seismic data are measured at a scale of tens of meters while well data at a scale of tens of centimeters. In addition, seismic data can be poorly processed; some well logs can be damaged, affected by mud filtrate invasion or completely missing. This research addresses the methods of (1) editing, conditioning and petrophysical analysis of well logs and (2) joint calibration of seismic and well data to improve correlation and consistency between the two data types. A case study using a data set from the Gullfaks filed is presented; this field is in tail production and therefore improved seismic reservoir characterization to prolong its production life is quite essential. With the help of Geoview, Elog and AVO modules of Hampson-Russell software and Geovation/Geocluster software; post-stack processing, petrophysical modeling and analysis, and joint-calibration of the data were carried out. The results show that locally calibrated rock physics models (of for instance Gardner's and Castagna's equations) produce more accurate synthetic well logs (of missing or damaged curves) than those produced using Global' relations. Fluid replacement modeling was carried out to factor in the presence of hydrocarbons in the reservoir zones; the results show more accurate prediction of well logs in the reservoir zones. The quality of well logs was greatly enhanced, in preparation for the joint calibration process. Multi-well wavelet extraction and analysis was done to extract a single wavelet; the wavelet so extracted produced synthetic data that correlates well at all well locations. In some of the wells the correlation coefficient was over 0.50. In one of the wells the correlation coefficient rose from -0.40 (for an individually extracted wavelet) to 0.30 (using a multi-well extracted wavelet). The study demonstrates that it is possible to obtain a high correlation between seismic and well data, if the data are well processed and conditioned. Multi-well wavelet extraction produces a wavelet that is applicable at all well locations