Due to a sharp fall in oil prices in late 2014, many oil exploration companies have either stopped operations or postponed projects to a future date. The resulting slowdown has strengthened the dependency on existing developed fields for oil production. This is a cause of concern for major oil corporations and governments worldwide, as the dependence on mature fields suggests that conventional oil extraction techniques may not be enough to maintain current demand and may lead to significant profit losses. Thus, the development of enhanced oil recovery (EOR) (also known as tertiary recovery) methods to improve recovery from developed fields has attracted attention.
Thermal recovery, a widely used EOR method in heavy oil reservoirs, involves the introduction of heat into the formation to reduce the viscosity of the oil in the reservoir. Cyclic steam stimulation (CSS) is an effective thermal process used with naturally fractured reservoirs. The cyclic steam injection (CSI) method incorporates the stages of injecting, soaking and production one by one in a single well.
The use of a commercial simulator for estimating production is common. However, the process can be time consuming and complex. Alternatively, it is possible to achieve results within seconds using an adequately trained artificial neural network (ANN).
This study analyzes CSI performance based on its effectiveness with respect to viscosity contours and cumulative oil production. Naturally fractured reservoirs are excellent targets for steam injection because they possess a structure where steam can easily diffuse.