Flooded Layer Log Interpretation And Distribution Of Remaining Oil

The process of waterflooding in oil fields becomes very complicated as a result of the anisotropy and heterogeneity of the reservoir in horizontal and vertical. With the oil layer entering into high waterflooding stage, its physical property, lithologic and oil saturation have changed and due to the various well logging response, It makes log interpretation more difficult. In order to solve this problem, besides developing well logging methods, the study of new log interpretation technique is of great importance.In view of this condition, this dissertation focuses on log interpretation methods in high water cut stage, aiming at the remaining oil distribution and reservoir evaluation, according to the remaining oil distribution in high water cut oilfields and the main task during the stage of deep exploitation, the variation of the reservoir physical property and lithologic in the process of water flooding have been systemically studied. On the basis of the petrophysic experiments' results, correlation analysis method, artificial neural network technique, fuzzy statistical method, stochastic simulation technique are applied to establish well logging interpretation model and the systematic research approach on prediction of the remaining oil distribution, which is a new way to predict the remaining oil distribution in high water cut stage. The main results of this paper are as follows:(l)The physical property variation rules of water flooded layerBased on the effects on water flooded layer caused by sedimentary microfacies, structure and production manner, the characteristic of reservoir litofacies, porosity, permeability, rock wettability, salinity, neural response and reservoir acoustic property have been analysed, and the physical property variation rules of water flooded layer and well logging responses are concluded, which is the basis of recognizing water flooded layer and establishing interpretational model.(2)The research into recognize the water flooded layer by using qualitative methods(1)Applying correlation analysis method to identify water flooded layer: Correlation analysis method is first proposed to identify water flooded layer in this paper. In this method, correlation coefficient of various kinds of layers such as oil layer, water layer and water flooded layer are qualified to distinguish which kind of layer it belongs to.(2)The basic principle of artificial BP neural network is studied and a new BP network isproposed. With improved BP network, Adopting deep, short resistivity, and acoustic logging, the model to recognize oil layer and water flooded layer is established.(3) Fuzzy statistical method is adopted to recognize the oil layers, water layers and water flooded layers. First, the fuzzy set that characterize oil layers, water layers and water flooded layers is established. Second, fuzzy closeness rules is applied to recognize reservoir type.(3) The study of log interpretation model of water flooded layers(4)On the basis of the results of core petrophysics experiments, with the mathematical and theoretical analysis of the "U" curve, the new interpretation model of water flooded layer based on the Archie equation is established. Simultaneously, the model' adaptability in the various condition is discussed.(5)C/O values of shaly sands matrix have been got from the neutron activation analysis, then the influence factors of shale on C/O value are determined.(6)On the basis of the theory on predicting remaining oil saturation with the application of productive logging, a new way that is used to determine remaining oil saturation and water rate is proposed by connecting relative permeability acquired by core analysis and productive data.(4) The simulation of distribution of interwell remaining oil(1)Connecting well logging interpretation with flowing tube method, a method that using remaining oil saturation of productive wells to predict distribution of interwell remaining oil is proposed.(2) The distribution of remaining oil is simulated by using Simulated annealing method. The results reflect the var...