Abstract:

Volcanic lava is widely distributed in Yingcheng Formation of Yingtai fault depression in southern Songliao Basin. It is one of the most important hydrocarbon reservoirs of volcanic reservoirs. The reservoir is characterized by large thickness, excellent reservoir properties and high productivity, and has favorable prospect. Volcanic lava has complicated lithologies, variable seismic reflection characteristics and strong heterogeneity, so it is difficult to define the volcanic lava boundary and identify volcanic apparatus. Besides, deep reservoir burial, lowquality of seismic data, low-angle of gathers and superposed wave impedance enhance the difficulty of the reservoir prediction. This paper applied several techniques to identify the boundary and inner of volcanic lava of Yingcheng Formation, and defined its distribution. According to the log response and gas reservoir characteristics, we used pre-stack simultaneous inversion technology based on log reconstruction, which improves the inversion accuracy and reliability of prediction of volcanic lava reservoir. This method provides newthought and method for the identification and prediction of this kind of complicated reservoirs.

Key words: tight sandstone, porosity, generalized regression neural network, heterogeneity, Chang 8 oil reservoir set, Jiyuan area