Study On Simultaneous Iterative Reconstruction Technique Inversion Method Of NMR Relaxation Signal

Since nuclear magnetic resonance phenomenon was discovered in1946, soon, ithas been widely used in chemical, physical, medical and biological fields. In1991,the Numar Company invented the nuclear magnetic resonance imaging logginginstrument. Use the CPMG pulse sequence to measure the NMR signal of hydrogennucleus in formation pore fluids, and get the transverse relaxation time T2distribution,then obtain the physical parameters of rocks, such as: the formation porosity, claybound water, capillary bound water and flow volume. Under certain conditions, weobtain the formation pore size distribution, permeability and fluid type saturation, andit provides important information for reservoir evaluation and productivity prediction.At the same time, it has been widely used in low resistance reservoir, lowpermeability reservoir and complex lithology reservoir of comprehensive evaluation,and solves a lot of problems in oil and gas exploration and development.The measuring initial data of NMR logging gets together of formation saturation,permeability, porosity and oil and gas information, it is a collection of all theinformation. But these we want information which cannot be directly from the initialdata analysis, so we should decode it, and the decoding process is unfolding.The fluid relaxation mechanism is mainly longitudinal relaxation T1andtransverse relaxation T2in NMR logging. Because longitudinal relaxation T1processis long, so in borehole measurement process mainly adopts transverse relaxation T2measurement. A single pore fluid relaxation process is a single-exponential decayprocess, but the observation object of NMR logging is multi-pore, so the inversionprocess is to solve a multi-exponential decay. By now, there are a lot of inversionmethods, such as: model smoothing, transform inversion method, conjugate gradientmethod, basis function method and single value decomposition, etc.This article is based on principles of NMR logging and the experience of theoriginal method. This paper uses a simultaneous iterative reconstruction techniquealgorithm (SIRT), it is an algorithm proposed by Dines and Lyttle to improve theconvergence of the ART algorithm. Comparing with other algorithms, the SIRTalgorithm is easier to be understood, easier to be implemented; it needs less artificial control which reduces the error caused by human factors. By combining with the SVDalgorithm, making the algorithm of convergence faster, comparing with SVDalgorithm, the computation time is faster than the latter. The SIRT algorithm is stableand inversion results is good continuity, especially in low SNR data inversion, resultshave a high resolution. The non-negative property of T2spectrum is much easier to beimplemented.In this paper, after data collection, algorithm designing, program compline,verification of the correctness of the program and the NMR logging data processingthese five steps. Through the comparative analysis with Halliburton Petrosite softwareprocessing, the processing results of the program is consistent to that of Petrositesoftware. Especially for low SNR, the result is better than the Petrosite software, cansuppress noise single better.