Abstract Experimental and numerical studies were made of water coning in an oil-producing well under two-phase, immiscible, incompressible flow. The model chosen was a pie-shaped, cylindrical sand pack with radial symmetry. Saturations were measured in situ by 70 micro-resistivity probes embedded in the sand pack. Results indicated that the numerical model pack. Results indicated that the numerical model simulated the experiments adequately. Increasing the production rate or the wellbore penetration lead to earlier water breakthrough; however, oil recovery was independent of production rate. As the ratio of gravity to viscous forces increased, the oil recovery at any given WOR became greater; wells should have been spaced closer if the horizontal permeability was low or if the vertical permeability permeability was low or if the vertical permeability was high. High vertical permeability decreased the oil recovery, while the opposite was true for horizontal permeability. In stratified formations, the highest permeability. In stratified formations, the highest oil recovery resulted when the most permeable section was located near the top of the oil-bearing zone. Introduction Coning in oil-producing wells is a problem more common than generally is believed. It occurs in producing formations that are underlain by water, producing formations that are underlain by water, overlain by gas, where a secondary gas cap develops, or are under the conditions of water, gas, or solvent injection. The present oil shortage has resulted in wells being produced at full capacity -- a situation that aggravates coning. Under severe coning conditions, well allowables must be reduced to. a level that minimizes coning and avoids loss of ultimate oil recovery. These considerations make study of the coning phenomenon more important than previously. previously.The two objectives a this study wereto apply a numerical coning model to actual laboratory results to verify validity of the numerical model, andto use the numerical model to study the effect of certain parameters on development of be cone and on the oil-recovery performance. The study was restricted to water coning in oil wells in a reservoir system of cylindrical geometry with radial symmetry. PROCEDURE PROCEDURE SELECTION OF A NUMERICAL MODEL The Blair and Weinaug problem was solved using four different numerical coning models and the solutions obtained were compared with the results of Letkeman and Ridings to select the numerical model to be used in the rest of the study. EXPERIMENTAL STUDY For the experimental study, a pie-shaped, cylindrical coning model was constructed of clear plexiglass. The model was 16 in. high and had a plexiglass. The model was 16 in. high and had a radius of 20 in. and an angle of 30 degrees. It was constructed of 1-in.-thick plexiglass and was supported by a metal frame all around to avoid bulging during flow. To distribute the injected water uniformly across the bottom face of the sand pack, the bottom plate had fluid grooves that were pack, the bottom plate had fluid grooves that were overlain by several layers of 325-mesh monel screen. Seventy micro-resistivity probes were constructed and positioned inside the model for measuring the electrical resistivity. Probes were constructed from two 1/8-in.-square, 100-mesh monel screens positioned parallel to one another and separated by positioned parallel to one another and separated by about 1/4 in. A thin, insulated wire led from each screen through a hole drilled in the side plates of the model to an electrical 70-point junction box, and from there to a specially constructed 70-channel scanner. The scanner could scan any or all of the resistivity probes at a rate ranging from 112 to 60 seconds per probe. A timer permitted continuous scanning or time-lapse scanning ranging from once per hour to once every 8 hours. The output of the per hour to once every 8 hours. The output of the scanner was put through a digital voltmeter to a digital recorder. During a flow experiment, the resistivity at each probe thus could be measured and recorded automatically. The resistivity probes were positioned. on a central symmetry plane, thus permitting measurements far from the boundaries of the model. SPEJ P. 247