Waterflood optimization using an injector producer pair recovery factor, a novel approach
AbstractThis paper presents an approach to optimize the recovery factor and sweep efficiency in a waterflooding process by automating the optimum injection rate calculations for water injectors using streamline simulation. A streamline simulator is an appropriate tool for modern waterflood management and can be used to determine the dynamic interaction between injector and producer pairs, which will vary over time based on sweep efficiency and operational changes. A streamline simulator can be used to identify injectors, which are not supporting production and contributing mainly to water producing wells. Streamlines illustrate natural fluid-flow paths in the reservoir, which are based on fluid properties, rock properties, well distribution and well rates across the reservoir. A bundle of connected streamlines can provide the oil in place between an injector/producer pair at any given time during a simulation run. Thus, the well pair recovery factors for each injector/producer pair, the produced water cut and the weighting factor for each injector are determined. Multiplying this weighting factor by the injection rates determines the new injection rate for each injector. For a well pair water cut that is lower than the average field water cut, the injection rate will be increased and vice versa. Given a finite volume of injection water, there will be a re-allocating of water from a well pair with a low recovery factor and high water cut and redistributing the water to injectors supporting low water cut producers, thus maximizing the recovery factor and reducing the field water production. The described approach is an automated procedure during the reservoir simulation run, making it appropriate for full field waterflood optimization with many injectors and producers in high-resolution heterogeneous brown reservoirs. This approach can reduce the water cut and increase the recovery factor and extend the life of the waterflooded oil fields. It was initially tested with a synthetic model and later with an actual reservoir model, which will be described in this paper.