Waterflooding is less effective at expanding reservoir production due to interwell thief zones. The thief zones may form during high water cut periods in the case of interconnected injectors and producers or lead to a total loss of injector fluid. We propose to identify the thief zone by using a support vector machine method. Considering the geological factors and development factors of the formation of the thief zone, the signal-to-noise ratio and correlation analysis method were used to select the relevant evaluation indices of the thief zone. The selected evaluation indices of the thief zone were taken as the input of the support vector machine model, and the corresponding recognition results of the thief zone were taken as the output of the support vector machine model. Through the training and learning of sample sets, the response relationship between thief zone and evaluation indices was determined. This method was used to identify 82 well groups in M oilfield, and the identification results were verified by a tracer monitoring method. The total identification accuracy was 89.02%, the positive sample identification accuracy was 92%, and the negative sample identification accuracy was 84.375%. The identification method easily obtains data, is easy to operate, has high identification accuracy, and can provide certain reference value for the formulation of profile control and water shutoff schemes in high water cut periods of oil reservoirs.
Using a support vector machine method to predict the development indices of very high water cut oilfields
