Multi-parameter Reservoir Fluid Identification Methods

At present,with the AVO (Amplitude Variation with Offset) technologyuncreasing development and perfection,especially prestack simultaneousthree-parameter (density,p-wave velocity,s-wave velocity) inversion technology ofmature development,three-parameter profiles can directly utilize the physicalrelationship of rock fluid properties to calculate the other members of the fluidattribute set,due to the parameters of fluid identification is more,and the effect ofeach parameter identification is different,so when choosing fluid identificationparameters,how to choose hypersensitivity parameters is one of the key steps toreservoir prediction.This paper introduced the basic theory of petrophysics and elastic parameters,studied the propagation law of seismic wave in a two-phase medium,expounded twokinds of effective medium theory.Then we first calculate fluid identificationcoefficient of elastic parameters or attribute parameters by using quantitative crossplot technique.According to the fluid identification coefficient to selectmulti-parameter,we get hypersensitivity parameters to use for fluididentification.Then,we use neural network pattern recognition technology to identifythe fluids.Finally,we use Castagna and Smith model and theoretical model to identifythe fluids.Due to the parameters extracted from seismic data is more,we reference to akind of quantitative rendezvous technical to select the sensitive parameters and weuse these hypersensitivity parameters to do fluid identification.What we do is notonly to avoid the use of low sensitivity parameters for fluid identification andreservoir prediction causing the uncertainty and multiplicity,but also greatly improvethe efficiency of the multi-parameter fluid identification.To do the fluididentification analysis,we can see,one-dimensional and two-dimensional intersectioncan be more clearly to separate the target reservoir,especially the three-dimensionalintersection can better distinguish the reservoir.Basing on BP neural network patternrecognition technology,we take advantage of sensitive parameters which we chosebefore by quantitative cross plot technique,as net input into the neural network set,comparing and analyzing the recognition effect the velocity and density as netinput.This can highlight the advantages of sensitive parameters as the input set.Finally,basing on theoretical research and analysis,we closely combined with oilfield production and practice.We make use of some wells in shengli oilfield area datafor the studying,using cross plots to verify at first and identify the reservoir fluidtype.Then with the BP neural network pattern recognition technology identify theunknown area of reservoir fluid.