| Nuclear magnetic resonance(NMR)logging is an effective means for reservoir evaluation,which has been widely used in the exploration and development of oil and gas resources.In traditional NMR logging applications,it is necessary to get NMR spectrum from echo data by inversion first,and then the required reservoir information can be extracted from the inversion spectrum.Reservoir parameter evaluation,capillary pressure curve prediction and characterization of bound and movable fluid microdistribution are three important aspects of NMR logging applications.The existing NMR logging application methods face the following problems: the NMR logging application is affected by the uncertainty of echo data inversion;the models used in NMR logging application need to be further optimized to meet the needs of unconventional reservoir evaluation;the existing inversion methods have large errors for echo data with low signal-to-noise ratio(SNR);there is no efficient method for the continuous and quantitative characterization of bound and movable fluid microdistribution.For reservoir parameter evaluation,three time domain analysis methods are proposed without inversion of NMR echo data to obtain transverse relaxation time(T2)spectrum,which can avoid the influence of inversion uncertainty on the bound water saturation and permeability evaluation.The first method is proposed for bound water saturation and permeability evaluation based on the calibration of NMR echo data.The relationship between bound water saturation and echo data is established according to the tapered cut-off value theory and the kernel function of echo data,and then the evaluation model of bound water saturation and permeability is established based on the optimal window value of echo data.The second method is proposed for bound water saturation and permeability evaluation based on exponential hyperbolic sine transform.A new method is proposed for construction of the tapered cut-off value function based on the exponential hyperbolic sine function,and then the bound water saturation is directly estimated from echo data by exponential hyperbolic sine transform.The determined tapered cut-off value function is added to the T2 arithmetic mean as a weight to construct the new weighted T2 arithmetic mean value(T2wam),which can better reflect the pore flow capacity.The integral transform method is used to extract the T2 wam directly from echo data,and the new permeability evaluation model is established based on T2 wam.The third method is proposed for permeability evaluation based on variable characteristic parameters in time domain.Three kinds of permeability factors indicating the flow capacity of rocks are directly extracted from echo data by integral transforms,and the form of each type of permeability factor is controlled by the parameter n.Then a new permeability evaluation model is established based on these permeability factors.The processing results of numerical simulation,core experiments and logging data verify the effectiveness of the three time domain analysis methods proposed in this paper for reservoir parameter evaluation.A new method for capillary pressure curve prediction using integral transform,the quantum genetic algorithm,and the artificial neural network(IT-QGA-ANN)based on NMR echo data is proposed.First,7 parameters that characterize rock properties are directly extracted from echo data using integral transforms.This process does not require the inversion of the echo data to obtain the T2 distribution,thereby avoiding the influence of inversion uncertainty.These characteristic parameters are then used as the input of the ANN,and the weights and thresholds of the ANN are optimized by the QGA method to improve its prediction ability.The processing results of core experiments and logging data indicate that,compared with the inversion-conventional nonlinear regression,IT-conventional nonlinear regression and IT-ANN methods,the proposed IT-QGA-ANN method can more accurately predict the capillary pressure curve.The NMR logging applications,such as fluid identification and microdistribution characterization,still rely on the spectrum obtained by NMR echo data inversion.Therefore,it is necessary to develop accurate NMR inversion methods.A method of NMR inversion constrained by general prior information is proposed.Different kinds of prior information are extracted from echo data by integral transforms,and then added into the residual term in the objective function of L2 regularization inversion method as the constraints to reduce the uncertainty of the solution.The proposed method improves the one-dimensional and multi-dimensional accuracy of NMR inversion based on the optimal combination of different prior information.A method is proposed for the continuous and quantitative characterization of bound and movable fluid microdistribution in porous rocks based on NMR T2 domain analysis.Based on the NMR experiments of conventional sandstone and tight sandstone samples,the microdistribution laws of bound and movable fluid in sandstones are analyzed.Then,the exponential function based on the natural constant is used to quantitatively characterize the proportional distribution of bound and movable fluids in rock pores with different sizes.Two characterization models of bound and movable fluids microdistribution are established according to whether the T2 distribution of the saturated fluid contains short relaxation peaks or not.The effectiveness of the fluid microdistribution characterization method is verified by core experiment and NMR logging data processing results. |
