论文部分内容阅读
在阿拉斯加Prudhoe Bay油藏的一组建议注水的模拟中进行了重力正演、反演模拟。一种通过地面重力观测未监测8200ft(2500m)深处储层中气顶注水进程的新监测技术已开发出来。这种经济有效的方法要求在数年内重复进行高精度的重力勘探。重力场随时间的变化反映出储层内流体密度的变化。在Prudhoe Bay油田进行的初次现场试验表明:把改进的Lacoste&Romberg“G”型重力仪或Scintrex CG-3M型重力仪和NAVSTAR全球定位系统(GPS)联合使用,勘探精度可达5μGal～10μGal正演重力模拟预测注水5年后地表测量的变化量为100μGal,15年后为180μGal～250μGal。我们采用约束最小二乘法对合成重力数据进行反演,以得到地下密度分布。模拟步骤已写成公式并已编成程序以便于测试模型对重力采样模式、噪声类型以及对各种模型参数(如密度、总质量和惯性矩)的约束的灵敏度。对于具有5μGal标准偏差(SD)噪声的合成重力数据的约束反演模型,其注水的水平特征分辨率大约为5000ft(1520m)。反演方法可以计算注入的总水量,误差在几个百分点之内受高级别的空间相关噪声(大于10μ～15μGal SD)干扰的重力数据的反演可导致最糟糕的结果在这种情况下根据反演模型估算的总质量可能会将质量高估或低估10%～20%。模拟结果表明时间推移重力数据的反演对于监测储层气顶注水是一项可行的技术。
Gravity forward modeling and inversion modeling were performed in a set of proposed waterflood simulations in Prudhoe Bay, Alaska. A new monitoring technique has been developed for observing the process of gas injection into the reservoir at a depth of 8200 ft (2500 m) by gravimetry. This cost-effective method requires repeated high-precision gravity surveys within a few years. Changes in gravity field over time reflect changes in fluid density within the reservoir. Initial field tests in the Prudhoe Bay field have shown that the combination of a modified Lacoste & Romberg “G ” gravity meter or a Scintrex CG-3M gravimeter and a NAVSTAR global positioning system (GPS) provides exploration accuracy of 5 to 10 [mu] Gal Gravity simulation predicts that the change in surface measurements after 5 years of flooding will be 100 μGal and 180 μGal ~ 250 μGal after 15 years. We use the constrained least-squares method to invert the composite gravity data to obtain the underground density distribution. The simulation step has been written as a formula and has been programmed to test the sensitivity of the model to gravity sampling modes, noise types, and constraints on various model parameters such as density, mass, and moment of inertia. For a constrained inversion model of synthetic gravity data with 5μGal standard deviation (SD) noise, the horizontal feature resolution of water injection is approximately 5000 ft (1520 m). The inversion method can calculate the total amount of injected water with an error of a few percent. Retrieval of gravity data that is disturbed by high levels of spatially correlated noise (> 10μ ~ 15μGal SD) can lead to the worst outcome under this scenario The estimated total mass of the inversion model may overestimate or underestimate the mass by 10% to 20%. The simulation results show that the inversion of gravitational data over time is a feasible technique for monitoring gas injection on the reservoir roof.