【摘要】 苏里格气田开发目的层主要为下二叠统石盒子组8段及山西组1段致密砂岩储层。受河流相沉积控制,河道砂体纵向上多期叠置,横向上复合连片分布,地震储层预测精度有限,主要原因有:①储层的复杂性、强非均质性导致储层预测存在一定难度;②砂、泥岩纵横波速度差异小,限制了叠后地震储层预测的准确性;③地震资料品质差、横波速度预测不准确,降低了储层预测的精度。针对这些问题进行了储层预测方法优化。研究结果表明:①采用超道集、拉东变换、无速度剩余时差校正可以提高地震资料品质;②在岩石物理横波速度预测理论的基础上,采用多参数神经网络方法可以提高横波预测精度;③通过弹性参数敏感性分析,优选出纵横波速度比、泊松比、体积模量/剪切模量等敏感参数,并确定含气储层的弹性参数下限,在此基础上采用叠前反演技术进行含气储层预测,为区块富集区优选与井位部署提供了可靠依据。更多还原

【Abstract】 In Sulige gasfield, the tight sandstone reservoirs of both Lower Permian Shihezi 8 Member and Shanxi 1 Member, which are further influenced by fluvial facies deposition, are the main development targets. Channel sandbodies are characterized by vertical multi-cycle overlap and lateral massive distribution. And the precision of seismic-based reservoir prediction is limited resulting from three factors, including both complexity and strong heterogeneity making a great difficulty to predict reservoirs, smaller P-and S-wave velocity difference between sandstone and mudstone restricting post-stack prediction precision, and poor-quality seismic data and inaccurate prediction of S-wave velocity also reducing the precision.So, some methods to predict reservoirs were optimized. Results show that(1) to adopt super gather, Radon transform, and speed-free residual time difference correction can improve seismic-data quality;(2) based on theory of S-wave velocity prediction, one method of multi-parameter neural network can improve S-wave prediction accuracy; and(3) by means of sensitivity analysis on elastic parameters, P-wave velocity to S-wave velocity ratio, Poisson’s ratio, and K/μ are selected, and the lower limits of parameters are determined for gas-bearing reservoirs. And these reservoirs can be predicted by means of pre-stack inversion technique, so as to provide reliable basis for optimizing accumulation blocks and deploying well locations.更多还原