SONIC LOG PREDICTION USING MULTI-ATTRIBUTE TRANSFORM OF PETROPHYSICAL LOGS

Sonic logs are important pieces for effective quantitative reservoir characterization studies. Unfortunately, in many development wells, sonic logs are not readily available due to cost-effectiveness in the operation. In this paper, we analyze the relationship between measured P-wave velocity log and other measured petrophysical log data at two development wells in Anadarko Basin, Oklahoma. The aim of this study is to predict sonic log at five other development wells in the same field that will be use later for quantitative reservoir characterization study in the area.

The observed log data in this study include gamma ray, density, resistivity and neutron porosity. We derived a statistical relationship through linear regression analysis for each petrophysical logs. The derived empirical relationship between sonic and other logs is then applied at a blind well test location in the same field. The calibration at blind well shows that sonic log model predicted using empirical relationship from neutron porosity log has the highest coefficient of correlation with the recorded sonic log.

We also used multi-attribute transform using the previous four petrophysical logs to improve the correlation of the sonic log prediction model. The multi-attribute analysis were performed at each individual rock formations available at depth and joined later as one complete sonic log. The correlation coefficient of sonic log model predicted using multi-attribute analysis is higher compared to sonic log predicted using linear regression. Well-to-seismic tie were performed at well locations using predicted P-wave velocity logs to construct the synthetic seismics. Both synthetic seismics generated using P-wave velocity from multi-attribute transform and linear regression analysis were compared and tied to the seismic. Highest correlation between synthetic traces and seismic is achieved using using P-wave log generated from multi-attribute analysis.