A method to deduce two-phase mobilities from an array of resistivity logs with different depths of investigation has been proposed (Ramakrishnan and Wilkinson, 1996a). The basis for this interpretation method is that invasion is a downhole flow experiment. The resistivity profile from this invasion process contains information that can be inverted to predict how the well would perform under production, provided the displacement process is similar to that of invasion. The inversion procedure yields fractional-flow curves, filtrate loss, and residual and movable saturations at every depth. To facilitate comparison between production data and the interpretation, we present a two-phase flow model, in which the fractional-flow curves obtained at every depth are converted to cumulative oil and water flow rates inside the wellbore. The model assumes a commingled system, with a constant pressure drive. The numerical algorithm to compute the phase flow rates in the wellbore can be run in two different modes:
when a single-phase permeability is available at every depth and
when the measured total (oil+water) rate is used as an input.
The algorithm is flexible enough to accommodate the low reliability of logs across shoulder beds. An arbitrary combination of layers in which the log-based interpretation may be superseded is available to the user. We discuss the application of the flow-based interpretation method in a Middle-East well, in which the AIT Array Induction Imager Tool and the MicroSFL tools were run. With a known petrophysical relationship for the field, the log data were processed to deduce the two-phase flow characteristics. The processing yielded the hydrocarbon and water intervals, and the permeability barriers successfully. The water-producing interval was identified to be due to seawater encroachment. Because the well was produced with an openhole completion, we could directly compare the production log-based oil and water flow measurements with the model predictions. The production log algorithm was run with both of the above-mentioned modes. Except for some minor anomalies, the agreement between the predictions and the measurements is good. Finally, we show how the laboratory core fractional flow data and the log-processed results match one other.
