The prediction method makes use of flow capacity distributions in a heterogeneous reservoir.

A number of tests using varying degrees of data contamination shows that the model has the potential to reduce bad data effects. It was also shown that the model can be used to easily determine the reservoir floodable volume, the water injection schedule for secondary recovery operations, and an overall waterflooding performance.

Our current capability to accurately predict the performance of a reservoir given a detailed description of its heterogeneities, calls for an urgent need for an efficient method of describing these non-conformities at any given locations in a reservoir.

In this study, a suitable approximation algorithm was developed for use in the estimation of reservoir performance prior to waterflooding operations. This algorithm, a two dimensional cubic spline, constructs a smooth, and continuous function of the given data values. This smooth function with continuous first and second derivatives, removes the 'wiggly' and undulating characteristics often present in most polynomial approximations.