一种基于组分补偿的二维核磁共振测井数据高精度处理方法

二维核磁共振技术能够对储层中各类含氢流体进行无损、快速、定量的测量和表征，但受限于采集方式和参数，核磁共振设备在对页岩油等致密储层中的有机质、沥青等超快弛豫组分进行检测时，经常出现由于信号采集不完整所导致的二维谱中流体组分缺失或不准的问题.本文提出了基于超快弛豫组分补偿技术的T₂-T₁二维谱高精度反演方法，该方法将一维核磁共振前端信号补偿技术进行推广，通过在二维核磁数据反演前对回波数据进行组分补偿，能够有效解决二维核磁共振测井前端信号漏失的问题.实验及测井数据的应用表明，该方法在页岩油等富含快弛豫组分信号的储层中，可以得到更加精准和完整的储层信息.     

基于差分进化算法的核磁共振T2谱多指数反演

提出了一种基于差分进化(DE)算法的核磁共振弛豫信号多指数反演新方法.将核磁共振T2谱反演问题转化为带非负约束的非线性优化问题,不需要预先给定横向弛豫时间T2分布,直接利用差分进化算法进行反演计算.在测量信号低信噪比的情况下,计算机模拟和实验数据反演鄙表明了该方法在分析处理NMR弛豫信号中的有效性.     

核磁共振二维谱反演

核磁共振二维谱包含扩散系数 D 和横向弛豫时间 T₂ 的信息,利用核磁共振仪来测量多孔介质中物质的信息,根据其弛豫时间和扩散系数的差别来区分不同物质;利用全局反演方法,提出了核磁共振二维谱反演的物理模型和数学模型;介绍了传统的奇异值分解（SVD）和改进的奇异值分解反演算法;采用改进的奇异值分解法对核磁共振二维谱进行反演,其反演算法具有计算速度快和二维谱分布连续等优点. 它适合于信噪比较高的数据反演,当原始数据信噪比SNR≥100时,可以对二维谱图进行定量分析;当60≤SNR<100时,可以对二维谱图进行定性分析. 核磁共振二维谱可以一次性直接区分油和水,为核磁共振测井提供了新的科学技术.     

核磁共振测井TDA识别油气的影响因素

TDA（时域分析）方法是核磁共振测井的主要方法之一,在判断储层流体类型、性质和孔隙度计算方面有很大的优势,但在应用中却存在一定的局限性. 针对TDA方法在流体识别中存在的一些问题,通过数值模拟不同条件的地层对其在油气识别中的影响因素及适用性进行分析. 研究结果表明：TDA主要用来识别轻烃,采用D9TW采集模式采集核磁自旋回波串,TDA能识别的轻质油粘度<9 mPa.s;长极化时间不变,短极化时间T_WS提高到2 s,TDA能识别的轻质油粘度<4 mPa.s. 双T_W回波串的差（差值信号,转换为孔隙度单位称为差分孔隙度）与地层孔隙度、含烃饱和度、含氢指数、极化函数等影响因素正相关. 对于低信噪比的核磁数据,综合各影响因素,发现烃的差分孔隙度要>1.5 p.u.,TDA方法才能有效识别油气和计算孔隙度.     

核磁共振测井受激回波校正方法研究

核磁共振测井中,受激回波导致CPMG序列中前2个回波与期望值存在较大差异,严重影响仪器信噪比和储层评价效果. 为进一步改善核磁共振测井仪性能,通过探究受激回波产生机理,揭示其对CPMG回波串作用规律,提出了受激回波的校正方法. 同时针对仪器负载不同,相同发射功率产生射频场强度各异的特性,建立受激回波校正系数与射频场强的数学关系,在不同应用环境中实现对CPMG测量结果的校正. 实验表明,在信噪比较低的应用环境中,采用传统的舍去前2个回波的方法误差达24.58%,而该文设计的校正算法误差仅为4.9%.     

基于差分进化算法的核磁共振T2谱多指数反演

提出了一种基于差分进化(DE)算法的核磁共振弛豫信号多指数反演新方法. 将核磁共振T₂谱反演问题转化为带非负约束的非线性优化问题,不需要预先给定横向弛豫时间T_{2分布,直接利用差分进化算法进行反演计算. 在测量信号低信噪比的情况下,计算机模拟和实验数据反演都表明了该方法在分析处理NMR弛豫信号中的有效性. 关键词： 核磁共振 多指数反演 差分进化 岩心分析}     

一种降低井下核磁共振振铃的新方法

核磁共振（NMR）测井仪在测量过程中,振铃噪声幅度一般很高,会影响回波信号的检测. 常用的交叉相位对脉冲序列（PAPS）虽然能有效地降低振铃影响,但由于振铃与频率有关,使PAPS叠加只能在同频率之间展开,大大降低了NMR测井的纵向分辨率. 利用回波信号和振铃噪声的相位特征,设计了一种脉冲序列,采用PAPS和回波间叠加相结合的方式降低振铃噪声. 研究表明,相对于PAPS技术,提高了测量效率,同时提高了回波串的信噪比和NMR测井的纵向分辨率.     

基于岩石重构图像的核磁共振响应模拟

能够反映孔隙介质内部真实复杂结构的CT图像经常被用于模拟研究中并能取得较好的结果. 但是,在很多情况下缺少三维CT图像或者其分辨率受限,此时,更易于获取的二维薄片图像被用于构建三维的孔隙结构图像. 假定孔隙介质各向同性,基于二维图像,使用多点统计方法可以构建岩石的三维孔隙结构. 基于数字图像,采用随机游走方法模拟得到核磁共振响应,然后由模拟得到的回波串反演得到T₂分布. 研究表明,岩石重构图像中的模拟结果与CT图像中的模拟结果有较好的一致性. 基于数字图像的核磁共振响应模拟为分析不同孔隙结构的核磁共振响应提供了便利,同时,模拟结果也为验证孔隙结构的重构效果提供了依据.     

2D NMR技术在石油测井中的应用

近几年,2D NMR技术得到迅速发展,特别是在核磁共振测井领域. 该文将主要介绍2D NMR技术的脉冲序列、弛豫原理以及2D NMR技术在石油测井中应用. 2D NMR技术是在梯度场的作用下,利用一系列回波时间间隔不同的CPMG脉冲进行测量,利用二维的数学反演得到2D NMR. 2D NMR技术可以直接测量自扩散系数、弛豫时间、原油粘度、含油饱和度、可动水饱和度、孔隙度、渗透率等地层流体性质和岩石物性参数. 从2D NMR谱上,可以直观的区分油、气、水,判断储层润湿性,确定内部磁场梯度等. 2D NMR技术为识别流体类型提供了新方法.     

一种评价核磁共振测井仪探测特性的方法

核磁共振测井仪敏感区域的形状、探测深度以及信号强度是评价其探测特性的主要指标,而这几个主要指标完全由探头磁场分布所决定. 为了在核磁共振测井仪探头设计的前期,实现对探头探测特性的客观评价,该文从核磁共振的基本条件出发,利用有限元数值模拟方法分析了MRIL-P型核磁共振测井仪探头的磁场分布,并对其磁场结构进行了优化匹配. 根据核磁共振信号原理,并基于数值模拟得到的磁场优化匹配结果,进一步模拟得到MRIL-P型核磁共振测井仪敏感区分布,并验证了核磁信号强度与工作频率之间的变化关系. 结果显示：在梯度磁场中,核磁信号强度主要受磁场强度的影响,并随工作频率的增大而增大,数值模拟的结果与预测结果相一致;该方法可以用来在核磁共振测井仪探头前期的设计阶段评价其探测特性,指导总体方案的设计.     

In situ fluid typing and quantification with 1D and 2D NMR logging

In situ nuclear magnetic resonance (NMR) fluid typing has recently gained momentum due to data acquisition and inversion algorithm enhancement of NMR logging tools. T(2) distributions derived from NMR logging contain information on bulk fluids and pore size distributions. However, the accuracy of fluid typing is greatly overshadowed by the overlap between T(2) peaks arising from different fluids with similar apparent T(2) relaxation times. Nevertheless, the shapes of T(2) distributions from different fluid components are often different and can be predetermined. Inversion with predetermined T(2) distributions allows us to perform fluid component decomposition to yield individual fluid volume ratios. Another effective method for in situ fluid typing is two-dimensional (2D) NMR logging, which results in proton population distribution as a function of T(2) relaxation time and fluid diffusion coefficient (or T(1) relaxation time). Since diffusion coefficients (or T(1) relaxation time) for different fluid components can be very different, it is relatively easy to separate oil (especially heavy oil) from water signal in a 2D NMR map and to perform accurate fluid typing. Combining NMR logging with resistivity and/or neutron/density logs provides a third method for in situ fluid typing. We shall describe these techniques with field examples.     

Tracking pore to pore exchange using relaxation exchange spectroscopy

We observe the movement of water over time between pores of differing sizes in Castlegate sandstone. To achieve this, we perform an NMR transverse relaxation exchange experiment for several mixing times. The resulting data are converted to 2D T2 distributions using a 2D inverse Laplace transform (ILT). We show for the first time that quantitative analysis of ILT distributions enables one to extract characteristic times for different pores sizes. This information is potentially useful for permeability determination as well as better understanding of exchange between specific pore subpopulations.     

Processing DOSY spectra using the regularized resolvent transform

A new method for processing diffusion ordered spectroscopy (DOSY) data is presented. This method, the regularized resolvent transform (iRRT-the i denoting the adaptation of the method to evaluate the inverse Laplace transform), is better than conventional processing techniques for generating 2D DOSY spectra using data that has poor chemical shift resolution. From the same data, it is possible to use the iRRT to generate 1D subspectra corresponding to different components of the sample mixture; these subspectra compare favorably to 1D spectra of the pure substances. Both the 2D spectra and the 1D subspectra offer a vast improvement over results generated using a conventional processing technique (non-linear least-squares fitting). Consequently, we present the iRRT as a stable and reliable tool for solving the inverse Laplace transform problem present in experiments such as DOSY.     

A novel improved method for analysis of 2D diffusion-relaxation data--2D PARAFAC-Laplace decomposition

This paper demonstrates how the multi-linear PARAFAC model can with advantage be used to decompose 2D diffusion-relaxation correlation NMR spectra prior to 2D-Laplace inversion to the T(2)-D domain. The decomposition is advantageous for better interpretation of the complex correlation maps as well as for the quantification of extracted T(2)-D components. To demonstrate the new method seventeen mixtures of wheat flour, starch, gluten, oil and water were prepared and measured with a 300 MHz nuclear magnetic resonance (NMR) spectrometer using a pulsed gradient stimulated echo (PGSTE) pulse sequence followed by a Carr-Purcell-Meiboom-Gill (CPMG) pulse echo train. By varying the gradient strength, 2D diffusion-relaxation data were recorded for each sample. From these double exponentially decaying relaxation data the PARAFAC algorithm extracted two unique diffusion-relaxation components, explaining 99.8% of the variation in the data set. These two components were subsequently transformed to the T(2)-D domain using 2D-inverse Laplace transformation and quantitatively assigned to the oil and water components of the samples. The oil component was one distinct distribution with peak intensity at D=3 x 10(-12) m(2) s(-1) and T(2)=180 ms. The water component consisted of two broad populations of water molecules with diffusion coefficients and relaxation times centered around correlation pairs: D=10(-9) m(2) s(-1), T(2)=10 ms and D=3 x 10(-13) m(2) s(-1), T(2)=13 ms. Small spurious peaks observed in the inverse Laplace transformation of original complex data were effectively filtered by the PARAFAC decomposition and thus considered artefacts from the complex Laplace transformation. The oil-to-water ratio determined by PARAFAC followed by 2D-Laplace inversion was perfectly correlated with known oil-to-water ratio of the samples. The new method of using PARAFAC prior to the 2D-Laplace inversion proved to have superior potential in analysis of diffusion-relaxation spectra, as it improves not only the interpretation, but also the quantification.     

Methods and limitations of NMR data inversion for fluid typing

We introduce two NMR inversion methods within the framework of 1D NMR to extract fluid saturations by varying echo spacing and wait time. The first method connects the T2 distribution of each fluid with the overall apparent T2 distribution using a shift matrix. Each fluid's saturation and T2 distribution are extracted by minimizing the difference between the model T2 distributions and measured apparent T2 distributions. The second method relates a model T2 distribution of each fluid with CPMG echo trains using a global evolution matrix that governs the evolution of magnetization under T1, T2 relaxation, and diffusion. These methods will be useful whenever data are not sufficient for 2D NMR inversion. They are also much faster than 2D for fluid typing. We also point out an inherent limitation associated with NMR inversion methods for fluid typing. Whenever there is singularity in the inversion matrix caused by similar behavior of model function for different fluids, most inversion algorithms remove the solution space associated with the singularity and choose a solution vector of the minimum length. This results in equal proportions of different fluids in the final answer. If prior knowledge such as saturation or T2 shape of the oil is available, there are several methods to tailor the solution to our desired outcome. However, if there is no prior knowledge available, such ambiguity always exists irregardless of the inversion schemes.     

Deconvolution of compartmental water diffusion coefficients in yeast-cell suspensions using combined T(1) and diffusion measurements

An NMR method is presented for measuring compartment-specific water diffusion coefficient (D) values. It uses relaxography, employing an extracellular contrast reagent (CR) to distinguish intracellular (IC) and extracellular (EC) (1)H(2)O signals by differences in their respective longitudinal (T(1)) relaxation times. A diffusion-weighted inversion-recovery spin-echo (DW-IRSE) pulse sequence was used to acquire IR data sets with systematically and independently varying inversion time (TI) and diffusion-attenuation gradient amplitude (g) values. Implementation of the DW-IRSE technique was demonstrated and validated using yeast cells suspended in 3 mM Gd-DTPA(2-) with a wet/dry mass ratio of 3.25:1.0. Two-dimensional (2D) NMR data were acquired at 2.0 T and analyzed using numerical inverse Laplace transformation (2D- and sequential 1D-ILT) and sequential exponential fitting to yield T(1) and water D values. All three methods gave substantial agreement. Exponential fitting, deemed the most accurate and time efficient, yielded T(1):D (relative contribution) values of 304 ms:0.023x10(-5) cm(2)/s (47%) and 65 ms:1.24x10(-5) cm(2)/s (53%) for the IC and EC components, respectively. The compartment-specific D values derived from direct biexponential fitting of diffusion-attenuation data were also in good agreement. Extension of the DW-IRSE method to in vivo models should provide valuable insights into compartment-specific water D changes in response to injury or disease. (c) 2002 Elsevier Science (USA).     

T(1)--T(2) correlation spectra obtained using a fast two-dimensional Laplace inversion

Spin relaxation is a sensitive probe of molecular structure and dynamics. Correlation of relaxation time constants, such as T(1) and T(2), conceptually similar to the conventional multidimensional spectroscopy, have been difficult to determine primarily due to the absense of an efficient multidimensional Laplace inversion program. We demonstrate the use of a novel computer algorithm for fast two-dimensional inverse Laplace transformation to obtain T(1)--T(2) correlation functions. The algorithm efficiently performs a least-squares fit on two-dimensional data with a nonnegativity constraint. We use a regularization method to find a balance between the residual fitting errors and the known noise amplitude, thus producing a result that is found to be stable in the presence of noise. This algorithm can be extended to include functional forms other than exponential kernels. We demonstrate the performance of the algorithm at different signal-to-noise ratios and with different T(1)--T(2) spectral characteristics using several brine-saturated rock samples.     

On the inversion of diffusion NMR data: Tikhonov regularization and optimal choice of the regularization parameter

The analysis of diffusion NMR data in terms of distributions of diffusion coefficients is hampered by the ill-posed nature of the required inverse Laplace transformation. Na?ve approaches such as multiexponential fitting or standard least-squares algorithms are numerically unstable and often fail. This paper updates the CONTIN approach of the application of Tikhonov regularization to stabilise this numerical inversion problem and demonstrates two methods for automatically choosing the optimal value of the regularization parameter. These approaches are computationally efficient and easy to implement using standard matrix algebra techniques. Example analyses are presenting using both synthetic data and experimental results of diffusion NMR studies on the azo-dye sunset yellow and some polymer molecular weight reference standards.     

Determination of Particle Size Distribution by the analysis of intensity-constrained multi-angle photon correlation spectroscopic data

Laser light scattering (LLS), especially dynamic laser light scattering (DLS), also known as photon correlation spectroscopy (PCS), is a well established method for particle size distribution analysis. It usually involves a Laplace inversion of the field autocorrelation function. However, the resolution is limited because of the ill-conditioned nature of this Laplace inversion. No unique solution exists when noise is present on the data. In contrast with this ill-conditioned nature, the angular dependence of scattered (static) intensities is precisely not ill-conditioned, which allows the resolution of the ill-conditioned inversion of DLS data to be improved. In order to characterize samples with more complicated size distributions, an intensityconstrained multi-angle PCS data analysis program has been developed, which is an alternative way of normalizing the field correlation function to that reported by Cummins and Staples [12]. In this program, the field autocorrelation function is normalized to the scattering intensity by using a predetermined coherent factor at each angle, which provides an additional constraint on the Laplace inversion of multi-angle PCS data analysis. The alternative analysis improves the resolution of PCS and provides a more reliable particle size distribution than single-angle data analysis. Both simulated and measured LLS data are used to illustrate its application, resolution and limitations.     

Optimal Filtering Applied to the Inversion of the Laplace Transform

Inversion of the Laplace transform, used in the laser scattering measurement of colloidal particle size distributions, presents severe numerical difficulties. In the presence of noise the variance of the inversion integral is infinite, indicating maximum uncertainty in the inversion. This paper applies the method of minimum variance, or “optimal”, filtering to the eigenfunction spectrum of the Laplace transform, giving an inversion which has finite variance. Spectral decomposition using the eigenfunctions of the Laplace transform gives a representation of the noise and desired signals analogous to the Fourier spectrum used in linear system theory. It is possible to obtain a filtered estimate of the unknown linewidth distribution. The requirement that the variance of this filtered estimate is minimum leads to a Wiener-Hopf integral equation defining the optimal filter. The results of this paper provide a basis of comparison of all methods of inversion of the Laplace transform, including the extensive literature of colloidal particle sizing by laser scattering or photon correlation.