Resistivity measurements are of fundamental importance for the calculation of oil saturation in potentially producing reservoirs. The combined measurement of shallow and deep resistivities enables the determination of the parameters Rt, Rxo and di. But in complex reservoirs we have difficulty in obtaining a confident reading of Rt, due to the low vertical resolution of deep reading tools. In laminated reservoirs, for example, the deep induction reading ILD can be interpreted erroneously with the belief that the measurement refers to one bed. This may be true for extreme case of thick beds, but more often is not. This problem can be partly resolved by enhancement of the vertical resolution of the deep reading log through comparison with the high resolution (shallow resistivity) log. One approach is to use a high resolution log where there is good correlation with the deep reading log. This correlation can be better evaluated if we apply a filter to the high resolution log such that theoretically the resultant log has the same vertical resolution as the low resolution log. However, this assumes that the vertical response of the high and low resolution tools are available, and in practice this is often not the case. In this study we attempt to demonstrate an alternative approach where the filter can be obtained from consideration of the frequency domain. The technique compares the spectral energy of the high and low resolution logs. It is shown that the vertical resolution depends fundamentally on the spectral energy of the actual log based on Parseval Theorem. Next a linear regression is applied to the filtered high resolution and low resolution logs and for each depth a minimisation routine is applied to determine the best correlation interval between the logs. Finally a correction factor is applied to each point on the low resolution log. This factor is considered by the correlation coefficients over the interval minimised for each point. The results obtained with induction logs are promising and the metodology shoud be aplyed on diferent logs techniques.
|
