The method discussed in this paper is to research the relationship between water-out degree of various layers at different waterflood stage and influential factors (well location, sandbody type, communicating states with surrounding injectors and water injection distance) based on water-out behavior analysis of various layers in inspection wells drilled at different period, and to evaluate water-out degree by severe, mediate and weak or no levels for each layer in any well by using fuzzy comprehensive evaluation method and neural network pattern recognition technology, then to further determine remaining oil distribution in each layer. Practical verification of other inspection well and separate layer testing data are conducted. The coincidence rate is high. This method has been successfully applied in remaining oil research of Lamadian oilfield at late period of high water cut stage.
P. 197
