苏5、桃7区块不同粒度碎屑岩测井识别方法

苏里格气田的苏5和桃7区块石盒子组及山西组储层为碎屑岩储层，但碎屑岩粒度变化大。在岩石成分差别不大的情况下，岩石粒度是造成电性特征差别大的原因之一，能用测井资料较准确地识别出岩石粒度，再按不同粒度碎屑岩建立储层评价模型，则能提高碎屑岩储层评价精度。以苏5和桃7区块取心井为目标，从常规测井资料中提取指示不同粒度碎屑岩的9个特征参数，经聚类分析后形成包括泥岩在内的粉砂岩、细砂岩、中砂岩、粗砂岩、砾岩共6类碎屑岩样本841个，再经非线性映射变换处理，得出不同粒度碎屑岩点在X-Y平面上大多呈不同半径的圆形分布特征；根据矢量点间距离最小原则对测井曲线进行分层，再输入测井特征参数和不同粒度碎屑岩样本，就能用模式识别方法从测井资料中识别出不同粒度的碎屑岩。3口样本井取心段岩性识别符合率在93%以上；对非样本井进行了岩性识别处理，与岩心分类结果比较，也有着较好的一致性。

LOGGING RECOGNITION METHODS FOR CLASTIC ROCKS WITH DIFFERENT GRANULARITIES IN BLOCKS SU-5 AND TAO-7

The reservoirs of Shihezi formation and Shanxi formation in blocks Su 5 and Tao 7 in Sulige gas field belong to clastic rock reservoirs with strong variations of clastic rock granularity. When the rock components are not much different, rock granularity is one of reasons resulting in large electrical character difference, hence, the rock granularity can be recognized by the wire logging data, then, the reservoir evaluation model can be built for the clastic rock with different granularities, accordingly, the evaluation precision for clastic rock reservoir can be improved. Taking the wells with core samples in blocks Su 5 and Tao 7 as research objective, this study extracted nine character parameters indicating different rock granularities with the conventional logging data, and formed 841 clastic rock samples including such 6 types as shale, siltstone, fine sandstone, moderate sandstone, coarse sandstone, and gravel by clustering analysis; then after these rock samples were performed by nonlinear mapping transform , it indicated that the clastic rocks with different granularities respond to circle distribution characteristics with different radius in X-Y plane; finally on the basis of vector distance least principle, this study interpreted the layers in the logging curves, and input logging characteristic parameters and the clastic rock samples with different granularities, in the end, a pattern recognition method was applied to recognize the clastic rock of different granularities by logging data. The consistent rate of lithology recognition is up to 93% for the core sections in 3 sample wells. Finally, the paper performed lithology recognition processing for the non sample wells, and their results are very consistent with the core classification results.