Multi-point Statistical Modeling Constrained by 3D Seismic Data Reducing of Uncertainty of Sand Body Prediction

Key words: 
• reservoir modeling /
• seismic data /
• logging data /
• multi-point statistical modeling /
• training image /
• curves for sand body probability making /

Abstract: A study of reservoir modeling is present with 3D seismic and logging data by applying of multi-point statistical modeling, and by the combination of seismic and logging data. The results indicate that the combination of seismic and logging data can reduce uncertainty of reservoir modeling. The study includes the quality control of seismic data, the combination of soft and hard data, the application of multi-point modeling, the making of training image, and the selection of curves for sand body probability making. At last, the result obtained by the combination of 3D seismic and logging data makes a contrast with the result by only logging data. Seismic shale content cross sections are used as a criterion for these analysis and contrast, which include 3 levels: ① contrast of distributions of sand and shale for upper part and low part of the reservoir, ② contrast of distributions of sand and shale for different wells and their neighboring areas, ③ contrast of among well logging sand prediction cross sections, and both well logging and seismic sand prediction cross sections, produced by different random seeds. As a result of this paper, multi-point statistical modeling constrained by 3D seismic data obviously increases reasonableness of sand prediction inter-wells, and reduces uncertainty of reservoir modeling.