Evaluation Of Reservoir Physical Properties And Hydrocarbon Typing By Nmr Technology

This thesis includes two main parts,the one concerns permeability estimation from NMR T2 spectrum by means of neural network technology,and the other one is the new log interpretation method based on NMR T2 spectrum.Currently,there are many traditional permeability calculation methods using NMR logs.Although these methods are applied successfully to some extent,they still have some draw'backs.For one primary reason,all NMR information was not used for permeability estimation.This research aims to explore a new permeability calculation method by integrating the BP neural network method and the T2 spectrum and core permeability data.Prior to the permeability calculation,the core data were calibrated for precious depth by well logs,and high-quality of core data is selected.At the same time,high resolution T2 spectrum was used.Based on numerous of experiments,fi-nally the BP neural network with 7 levels and 1601 nodes has been selected for study.Permeability training range is of 0.05-125mD.Results show that the permeability from neural network exactly matches core data in both amplitude and curve shape,the correlation coefficient is in the range of 0.92-0.99(0.94 in average).Using neural networks the function between core permeability and NMR T2 spectrum is compli-cated multiple-dimension and non-linear,this function is much more accurate than existing permeability models,it can be applied to accurate permeability estimation of reservoir,and is prospective to be used widely in the future.Currently,there are two traditional methods for identifying hydrocarbon typing by NMR independently:the differential spectrum and diffusion methods.Differential spectrum method is suitable for reservoir of light oil and diffusion technique is suita-ble for reservoir of middle viscosity oil.However,both methods cannot be combined together for hydrocarbon typing analysis.Because oil is a kind of mixture of mul-ti-components,using any one of the above two models will result in lower oil satura-tion.By analyzing standard T2 spectrum,I found that the key feature of T2 distribution of water.Then I build the water T2 distribution successfully.There are two keys to build water T2 distribution;one is finding water T2 peak,the other one is obtaining distribution function of water.When perform for hydrocarbon typing analysis,I firstly build water T2 distribution,and then subtract water distribution from measured T2 spectrum.As a result,the hydrocarbon signal and hydrocarbon saturation were finally obtained.The method can be used for evaluation of hydrocarbon reservoirs with wider range of viscosity.