Font Size: a A A

Study Of The Evaluation On Coal-Gas Reservoirs Based On The Picking-up Of Well-Logging Information

Posted on:2008-02-22Degree:MasterType:Thesis
Country:ChinaCandidate:B Y XuFull Text:PDF
GTID:2120360242456682Subject:Earth Exploration and Information Technology
Abstract/Summary:PDF Full Text Request
Guided by sequence stratigraphy theories, this article adopts wavelet transform, fractal dimensions, principal component analysis and clustering analysis to deal with well-logging data and fully dig geological information which implies in well-logging series to serve identification and evaluation of coal-gas reservoir. In this article, Matlab function "wden" is used to denoise well-logging data and the use of soft thresholds and comparison is also made between signals before and after denoisig. It shows that using sym8 wavelet, sln mode to denoise signals at level 5 can obtain good denoising effect. This article does some researches on the sequence division of Carbon-Permian section of Jiyang Depression, especially the upper section with good developed reservoir. Well-logging data are respectively dealt with continuous transform and discrete wavelet transform .Sequences and parasequences are divided according to Oscillating trend of wavelet coefficients at different levels. It shows that it is effective to use dimension 81 or the high frequency coefficients at level 4 to divide sequences units and it is also effective to use dimension 21 or 46 to divide sequences units. By dealing PCA data with discrete wavelet transform and applying it to the division of sequences and parasequences, we discover that the sequences units interface divided according to PCA data correspond with those divided according to traditional methods. The first PCA curves, which have good are quite similar to GR curves on displaying characteristics of sequences objectivity. This article uses hierarchical clustering, K-means clustering and fuzzy C-means clustering to recognize lithology and compare hierarchical clustering and K-means clustering. It reveals that K-means clustering is better than hierarchical clustering on recognizing lithology of complex clastic rock while hierarchical clustering is better than K-means clustering on recognizing coal. By analyzing the physical property values of sandstone in study area, the changing rules of physical properties of different formation can be obtained. We also find that porosity varies with depth and has something to do with permeability. Besides, AC fractal dimension curves, which can be considered as assistant indexes, have good correlativity with porosity curves, and their changing trends are similar to porosity curves. By using principal component analysis, we can get a series of proper vector values, which can be use to draw cobweb map. After comparison we obtain reservoir identification of Kuishan and Wanshan section.
Keywords/Search Tags:well-logging data, sequence strata, wavelet transform, fractal theory, principal component analysis, clustering analysis, wavelet denoising, division of sequences, identification of lithology, anisotropism of Reseroir
PDF Full Text Request
Related items