Font Size: a A A

Study On EMD Phased Logging Reservoir Prediction Based On Cloud Model

Posted on:2017-03-15Degree:DoctorType:Dissertation
Country:ChinaCandidate:N ZhaoFull Text:PDF
GTID:1220330488463463Subject:Earth Exploration and Information Technology
Abstract/Summary:PDF Full Text Request
In lithologic stratigraphic reservoir exploration, as low permeability reservoir is characterized by poorer physical properties, so logging curve SNR is lower, and reservoir fluid identification is more difficult, with multiple solutions. EMD phased logging reservoir prediction based on cloud model, according to advantageous logging phase sequence, intends to explore empirical mode decomposition(EMD), improve SNR, and then enhance the resolution of logging data; with layerbreak as the sample, it is a new method to study time-frequency analysis technology in logging sequence stratigraphy; on the basis of fine layering of major reservoir, it provides new algorithm and new parameters for reservoir fluid identification based on EMD decomposition; as for multiple solutions in reservoir productivity evaluation, it establishes 2D normal cloud model assessment criteria, and achieves visualized logging productivity evaluation process; through the analysis of low permeability tight sandstone reservoir, it gives logging interpretation results scientifically and objectively, in attempts to guide oil and gas exploration process. Specifically, main research contents and results are as follows:(1) Comprehensively discuss algorithm theory for logging data processing. With respect to time-frequency analysis technology, it analyzes traditional Fourier transform and wavelet transform, as well as advantages/disadvantages and limitations of EMD decomposition algorithm, and it indicates: Fourier transform is suitable for the analysis of stationary signals with fixed frequency; wavelet transform and EMD decomposition are more suitable for the analysis of nonlinear and non-stationary logging sequences. In the study on spatial data mining theory, it describes "Cloud Model" theory for mutual conversion between qualitative concepts and quantitative data in the uncertain artificial intelligence, discusses the significance of "cloud" three basic digital characteristics, provides the algorithms for the realization of "Normal Cloud" model generator, and lays theoretical foundation for multiple solutions in logging reservoir productivity evaluation.(2) It deeply discusses the application skills of multi-parameter radar chart in logging phase control data processing, prepares radar chart drawing programs, and provides users with convenient use environment. Take Xujiahe Team in Dayi Structure in Sichuan Basin for example, it describes three typical methods and operation procedures for radar chart in logging data processing; establishes radar chart standard templates for local lithology and fluid identification; it uses radar chart drawing program to effectively help data interpreters accurately achieve the separation sensitive logging indicators, identify the unknown well section lithology and preliminarily determine fluid properties of reservoir stratum.(3) It deeply analyzes three methods for sequence interface and cycle identification based on logging data, and also provides new understandings. Take L Well in North Oil Plant in Ordos Basin for example, by comparing theoretical basis, application methods and skills of conventional logging curve shape recognition, wavelet transform fine stratification and empirical mode decomposition cycle identification, it indicates: traditional logging curve shape recognition method needs too large workload, it is more difficult, and greatly subject to artificial factors; wavelet transform is a time-frequency local analysis method, and it generates low frequency approximation signal and high frequency detail signal through two complementary filters, corresponding to signal stabilization trend and jump point. According to these characteristics, we can accurately achieve small fine layering; in EMD decomposition function, each component contains complete information of the signal, and it reflects instantaneous frequency characteristics, so it is more suitable for the identification of different stratigraphic sedimentary cycle.(4) It provides new methods for the identification of logging stratigraphic cycle based on EMD decomposition; summarizes basic forms of five EMD decomposition curve response characteristics in various depositional environments; based on Lissajous figures, it indicates that smooth closed curves appeared on IMFs component in the depth domain are exactly the performance of stratigraphic sedimentary cycle; uses main value range of IMFs instantaneous frequency to deduce the identifiable cycle times, and proves the scientificity of new approaches. In combination with the example of Xiaoquan in the middle section of Xi Ao Xian-Fenggu structural belt in Sichuan Basin, it describes new methods and operation steps for EMD decomposition in logging stratigraphic cycle identification, and deeply discusses geological significance of proper intrinsic mode function under different scales, it indicates: carry out EMD decomposition of regional sensitive parameters, according to relativity principle, optimize IMFs components to reflect more stable information of stratigraphic cycle; in the study on sequence stratigraphy, the stratigraphic cycle is classified based on distribution time limit. Main value range of instantaneous frequency is used to deduce discernible depth range of logging data in the depth domain. Therefore, it can well identify different stratigraphic cycle characteristics.(5) It provides new algorithms and new parameters for EMD-based reservoir fluid identification. The empirical modal analysis method is used for low permeability tight sandstone gas reservoirs. Through empirical modal decomposition of the appropriate gas-sensitive parameters, it carries out multi-scale comparative analysis of correlation, selects IMFs component which can well reflect the changes of fluid nature as cross-plot, reveals local curve correlation and local linear correlation laws hidden microscopically of non-stationary logging signals, finds out new water level indication parameter Pw, gas reservoir indication parameter Pα and high-quality gas reservoir indication parameter HGR, and provides methods and calculation formulas for reservoir fluid discrimination according to new parameters.(6) A new law fluid identification was found suitable for tight sandstone reservoirs. Take gas reservoir identification in Aoxian Xujiahe Team in West Sichuan for example, it reveals local correlation rules hidden in the advantageous logging phase sequence signals, and verifies the feasibility of fluid identification of new indication parameters and new algorithms of water layer and gas layer in low permeability tight sandstone reservoir. It deeply describes new algorithms and steps for gas and water surface identification with EMD decomposition under the conditions that gas reservoirs can not be recognized with conventional methods, and verifies the effectiveness of new algorithms and new arguments. It indicates: in accordance with the correlation principle, in the preparation of cross-plot for empirical mode decomposition function of preferred gas sensitive parameter AC and CNL, natural gas reservoir appears partial curve correlation features, showing obvious arc. The better the reservoir properties are, the greater the arc curvature is. Whereas the water zone and dry layer appear partial linear correlation or can be approximated into linear correlation features.(7) It provides new methods for logging reservoir productivity evaluation based on cloud model. In terms of uncertainties in oil gas reservoir productivity evaluation process, as well as the demands for quantitative and qualitative variable conversion and mapping between logging data and expert viewpoints, it deeply analyzes digital characteristics of cloud model. Take Yanchang stratum in X Well in North Oil Plant in Ordos Basin for example, it combines oil test data, builds 2D normal cloud chart for reservoirs with different productivity, and provides methods for the identification of reservoir productivity qualitatively and quantitatively, thus achieves visual interpretation of data mining. It comprehensively studies the distribution features of reservoir data, retains uncertainties for concept description in assessment process to the maximum degree, establishes qualitative comparability, and objectively evaluates reservoir productivity, thus ensuring the credibility of logging interpretation.
Keywords/Search Tags:Empirical mode decomposition, logging phase, cloud model, logging reservoir prediction, sequence stratigraphy, productivity evaluation, low permeability tight sandstone reservoir
PDF Full Text Request
Related items