Study On Processing And Analysis Methods Of Nuclear Magnetic Resonance Logging Data

NMR(Nuclear magnetic resonance logging) is the most exciting technology inlogging field at the end of the last century.Hydrogen nuclear become the observationtarget of the NMR logging because of the high content, large nuclear magnetic and it’smaximum signal. Since the measured signal doesn’t affected by the solid skeleton, soit can provide some parameters such as porosity that unrelated to lithology, irreduciblewater saturation, free fluid saturation and permeability.so that overcoming the defectsof the conventional logging which based on the volume model, solving the problemsthat cannot be resolved by conventional methods in complex lithology and lowresistivity oil and gas reservoirs. NMR logging research the content of the free fluid inthe rock by using the initial amplitude of the spin echo train, study the properties ofthe fluid and the mode of occurrence by using relaxation time, So that requires toinversion of NMR log data.Studies found that the transverse relaxation of the magnetization intensity signalin the single pore attenuates meet the single index law.While in a complex poresystem, the total transverse relaxation process is the sum of a variety of single indexlaw of attenuations. Inversion methods basically have the following several:1、Thesimple single and double exponential fitting method. This method cannot meet therequirements of the delicate problem’s processing which means it has not theapplicability.2、The singular value decomposition method of the multi-exponentialfitting. Because the too large condition number of the matrix, we choose the smallersingular value and cutting the original matrix of rank, and the gained continuity of T2spectrum is not good.3、The nonnegative least squares method. When added thepenalty term, there appears a constant related to SNR and experience alafa.It increasesthe uncertainty of solution when choosing alafa empirically. This paper analyses thenon-negative restriction from the aspect of the vector space. Based on the singularvalue decomposition, it is solved by iterative algorithm. The complex matrixdecomposition process is replaced by a solving process of least squares. Truncatingthe singular values with a constant which is related with the SNR meet the request that the stable T2spectrum can be got under the condition of different SNR. Becausethe coordinate components are not lost in the process of the iteration of T2, thegained spectrum’s continuity is better.The differences of the fluid vertical and transverse relaxation times、the differenthydrogen density、the diffusion characteristics of a variety of fluids in Strata and theviscosity of the oil affect the measurement signal to varying degrees. The NMRlogging use these differences in the formation fluids to built difference spectroscopyand shift spectrum method interpretation model to quantify the presence of oil andgas.In this paper, we use new algorithm to write a program to process actual data andget corresponding echo curve fitting、T2spectrum. According to the NMR loginterpretation related formula, the geophysical parameters such as porosity、boundfluid porosity and free fluid porosity and permeability are calculated and theinterpretation results is very closed to the results given by the comprehensive loginterpretation.