A Production Estimation Algorithm in Commingled Oil Wells Based Kernel Partial Least Square

Abstract

Commingling is employed in the petroleum industry to enhance oil recovery and reduce costs. It is of great importance to monitor the production of each oil well oilfields. Nowadays, more and more oilfields use chromatographic fingerprint to estimate single-zone production allocation. In order to insure the efficiency and affectivity of the commingled oil well exploiting, the productivity contribution of every single layer must be acquainted. Kernel partial least squares (KPLS) is a promising regression method for tackling nonlinear systems because it can efficiently compute regression coefficients in high-dimensional feature spaces by means of nonlinear kernel functions. Unlike other nonlinear partial least squares (PLS) techniques KPLS does not entail any nonlinear optimization procedures and has a complexity similar to that of linear PLS. Using the technology of crude oil chromatography fingerprint, an algorithm for predicting productivity contribution based on KPLS is proposed. The validity of the method is proved by laboratory artificial experiments. The maximum absolute error of predicted and real proportion is less than 10%. The model can also be applied to other wells which are similar to those used in the experiment. The experiment results show the prediction model is feasible.