Investigating statistical techniques to infer interwell connectivity from production and injection rate fluctuations
Reservoir characterization is one of the most important factors in successful reservoir management. In water injection projects, a knowledge of reservoir heterogeneities and discontinuities is particularly important to maximize oil recovery. This research project presents a new technique to quantify communication between injection and production wells in a reservoir based on temporal fluctuations in rates. The technique combines a nonlinear signal processing model and multiple linear regression (MLR) to provide information about permeability trends and the presence of flow barriers. The method was tested in synthetic fields using rates generated by a numerical simulator and then applied to producing fields in Argentina, the North Sea, Texas, and Wyoming. Results indicate that the model coefficients (weights) between wells are consistent with the known geology and relative location between wells; they are independent of injection/production rates. The developed procedure provides parameters (time constants) that explicitly indicate the attenuation and time lag between injector and producer pairs. The new procedure allows for a better insight into the well-to-well connectivities for the fields than MLR. Complex geological conditions are often not easily identified using the weights and time constants values individually. However, combining both sets of parameters in certain representations enhances the inference about the geological features. The applications of the new representations to numerically simulated fields and then to real fields indicate that these representations are capable of identifying whether the connectivity of an injector-producer well pair is through fractures, a high-permeability layer, or through partially completed wells. The technique may produce negative weights for some well pairs. Because there is no physical explanation in waterfloods for negative weights, these are also investigated. The negative weights have at least three causes: changes in producer bottom hole pressure (BHP), ignoring the attenuation and time lag between injector and producer pairs, and correlation between the injection rates or collinearity. The technique was successfully extended to handle varying BHPs. This extension eliminates many of the nonphysical weights and enhances the well connectivity inference. Ridge regression, used to treat collinearity, improves the inference about well connectivity for the North Sea field.