3-D Geological Modeling for Tight Sand Gas Reservoir in Braided River Facies

Abstract

Sulige gasfield, a typical tight sandstone gasfield in China, is characterized with poor reservoir property, drastic sedimentary faices change, small scale of effective sand bodies and strong reservoir heterogeneity. Considering the poor applicability of conventional geological modeling, a new modeling method is put forward as “multi-stage constraints, hierarchical facies and multi-step models” to represent the dual reservoir structure of “effective sand bodies in normal sand bodies”. Based on the prior geological knowledge, GR field was inverted by seismic and logging data through neural network recognition technology. Taking rock facies data at well points as hard data and sandstone probability volume derived from GR inversion in inter-well zones as soft data, several training images are obtained according to the reservoir differences of various developing layers, then rock facies model was established by multipoint geo-statistics method. In view that braided-river sedimentary system has strong influence on development types, frequency and scale of sedimentary microfacies, sedimentary microfacies model was built controlled both by rock facies and braided-river system. Eventually, effective sand body model was built with both of the discrete and continuous modelling method by integrating sedimentary microfacies, effective sand body scale, and reservoir properties distribution. In this research, a series of models were set up like GR field inversions, rock facies, sedimentary microfacies, reservoir parameters and effective sands. GR field inversion ensures the quality and multiple sources of the analysis data, breaks traditional seismic resolution limits and clarifies geological meaning of predicted sands. The rock faices model is faithful to hard data at the well points and shows fluvial channel morphology well in inter-well zones. The effective sand bodies model has great consistence with statistical property and geological knowledge. The modeling method discussed, using geological constraints as far as possible, reduces data interpretation uncertainty and improves the model's reliability. In 1200 m × 1800 m well pattern, model accuracy is up to 73% from 46% of the traditional methods, increased by 27%. It is conclude that the new modeling method can provide a more reliable geological basis for gas reservoir development.

AAPG Datapages/Search and Discovery Article #90291 ©2017 AAPG Annual Convention and Exhibition, Houston, Texas, April 2-5, 2017