【摘要】 为了揭示二维核磁共振测井响应机理，建立可靠解释基础，解决复杂非均质地层中流体精准识别与定量评价的难题，采用以T₁—T₂核磁共振测井为例，融合Bloch方程求解与仪器空间灵敏度分布的二维正演模拟方法，构建从微观磁化演化到宏观测井响应的全流程物理建模体系，模拟不同含水饱和度、钻井液滤液侵入深度及信噪比等因素对测井结果影响的方法。研究结果表明：（1）该模型基于核磁共振探测器的三维有限元模型，能模拟静态磁场（B₀）和射频磁场（B₁）分布，结合SR-CPMG脉冲序列逐点求解Bloch方程，利用空间灵敏度函数加权敏感体积内各单元贡献；（2）模拟不同含水饱和度条件下的T₁—T₂谱响应，显示含水饱和度升高使谱峰向短弛豫时间区域偏移，模拟谱峰位置相对理论值平均偏差不超过3.5%；（3）信噪比对结果影响显著，信噪比从5提升到15时，谱峰位置误差由8.3%降至2.1%；（4）不同探测深度的模拟能有效识别钻井液滤液侵入区与原状地层，侵入区T₂谱呈单峰，原状地层呈双峰结构。结论认为，该二维核磁共振测井正演模拟框架在复杂地质条件下适应性与预测能力良好，为定量流体识别与高精度储层评价提供了有效技术支撑。更多还原

【Abstract】 To reveal the response mechanisms of two-dimensional nuclear magnetic resonance（NMR） logging, establish a reliable interpretation basis, and address the challenges of accurate fluid identification and quantitative evaluation in complex heterogeneous formations, this study takes T₁—T₂ NMR logging as an example and develops a two-dimensional forward modeling approach that integrates Bloch equation solutions with instrument spatial sensitivity distributions. A comprehensive physical modeling framework was constructed, spanning from microscopic magnetization evolution to macroscopic logging responses, to simulate the effects of various factors such as water saturation, drilling fluid filtrate invasion depth, and signal-to-noise ratio（SNR） on the logging results. The results demonstrate that:（1）Based on a three-dimensional finite element model of the NMR detector, the model can simulate the distributions of the static magnetic field（B₀） and the radio-frequency magnetic field（B₁）. By combining the saturation recovery carr-purcell-meiboom-gill（SR-CPMG） pulse sequence to solve the Bloch equations point-by-point and utilizing a spatial sensitivity function to weight the contributions from each unit within the sensitive volume, the model achieves high precision.（2）Simulations of T₁—T₂ spectral responses under different water saturation conditions show that increased water saturation shifts the spectral peaks towards shorter relaxation times, with an average deviation of the simulated peak positions from theoretical values not exceeding 3.5%.（3）SNR significantly impacts the results: when the SNR increases from 5 to 15, the error in spectral peak position decreases from 8.3% to 2.1%.（4）Simulations at different investigation depths can effectively distinguish the drilling fluid filtrate invasion zone from the virgin formation: the T₂ spectrum of the invasion zone exhibits a single peak, while the virgin formation shows a bimodal spectrum. In conclusion, the proposed two-dimensional NMR logging forward modeling framework exhibits strong adaptability and predictive capability under complex geological conditions, providing effective technical support for quantitative fluid identification and high-resolution reservoir evaluation.更多还原