Empirical modeling and simulation of edgewater cusping and coning



Journal Title

Journal ISSN

Volume Title


Texas A&M University


In many cases, it is important to predict water production performance of oil wells early in, or maybe before, their production life. In as much as oil field water is important for pressure maintenance purposes and displacement of oil towards the perforation of the producing well, excessive water production leads to increased cost. In the case when no provision is made, it represents a significant liability. The case considered here is a well producing from a monocline with an edge-water aquifer. Although such problems can be computed with reservoir simulation, the objective of this work was to develop an empirical method of making water production predictions. The reservoir model was described as a single well producing from the top of a monocline drainage block with water drive from an infinite-acting aquifer. During the reservoir simulation runs, water would cusp and cone into the well, increasing water production and decreasing oil production. A number of simulation runs were made, varying eleven model variables. Typical model variables include dip angle, formation thickness and production rate. For each run a modified Addington-style plot was made. The relationship between each model parameter and three graphical variables was used to develop the set of empirical correlations. The empirical correlations developed were integrated with some derived equations that relate important reservoir parameters and incorporated into a computer program. The developed correlations and program can be used to carry out sensitivity analysis to evaluate various scenarios at the early planning stages when available reservoir data are limited. This gives a quick and easy method for forecasting production performance with an active edge-water drive. Furthermore, the approach developed in the research can be applied to other water production problems in other fields/reservoirs. The developed program was validated and used to evaluate synthetic and field cases. Overall, a good match was achieved.