论文部分内容阅读
一维核磁共振(1D NMR)测井技术在流体识别中具有一定的局限性。二维核磁共振(2D NMR)测井能更多参数包括多孔介质中纵向弛豫时间(T1)和横向弛豫时间(T_2)。根据梯度场下二维核磁共振弛豫机理,研究二维核磁共振回波串模拟与反演,并提出了基于阻尼最小二乘LSQR方法和改进的截断奇异值分解法的混合反演算法。在梯度场下,根据多等待时间,模拟给定流体模型的一系列回波串,并利用混合算法反演合成的回波串,反演结果与给定的流体模型匹配较好。并利用此反演算法,对气水模型,轻质油水和稠油水模型进行了不同回波间隔、不同等待时间组的数值模拟实验。最后,系统考察了不同观测参数对反演结果和流体识别效果的影响。此外,还系统研究了信噪比对多种流体模型反演结果的影响。数值模拟结果表明,混合算法与优化的观测参数非常适用于气水模型和油水模型。
One-dimensional nuclear magnetic resonance (1D NMR) logging technology has some limitations in fluid identification. Two-dimensional nuclear magnetic resonance (2D NMR) logging can more parameters include longitudinal relaxation time (T1) and transverse relaxation time (T_2) in porous media. According to the two-dimensional NMR relaxation mechanism under gradient field, two-dimensional NMR echo train simulation and inversion are studied. A hybrid inversion algorithm based on damped least squares LSQR and improved truncated singular value decomposition is proposed. Under the gradient field, a series of echo trains of a given fluid model are simulated according to multiple waiting times, and the hybrid echo algorithm is used to invert the synthesized echo train. The result of the inversion matches well with the given fluid model. By using this inversion algorithm, numerical simulation experiments of gas-water model, light oil water and heavy oil water model with different echo interval and different waiting time were carried out. Finally, the effects of different observation parameters on the inversion results and the fluid identification effect are systematically investigated. In addition, the effects of signal-to-noise ratio on the inversion results of various fluid models are systematically studied. The numerical simulation results show that the hybrid algorithm and the optimized observation parameters are very suitable for the gas-water model and the oil-water model.