Sedimentary Facies Analysis Of Triassic Leikoupo Formation In Xiaoquan-Xinchang Trap, Western Sichuan Using Seismic Sedimentology Technology

Western Sichuan basin holds rich natural gas reserve as large as 1.21×1012m3 and yet the proven rate is low, which presents one of the most promising exploration areas in Sichuan Basin. A series of important exploration wells have been deployed in this area by Sinopec Group since 2006. Chuanke-1, Xinshen-1 and Penzhou-1 Wells tested high gas flow. The success of these three wells bear breakthrough in the marine origin petroleum exploration field of Western Sichuan and reveal bright exploration prospect. The key of gas exploration in this area is to look for high quality reservoirs. However, the reservoir distribution is controlled by complicated sedimentary environment and diagenesis. Therefore, sedimentary facies analysis is an indispensable link between the reservoir predictions and carbonate rock. Every drilling costs much more than usual drilling due to the high burial depth of marine stratum. So the well data we need to complete the sedimentary facies division of this vast region and collaboration from seismic data is insufficient. In contrast to sparse well data,3D seismic data cover large areas with high density. Changes in the lithology and fluids result in changes in seismic amplitude, shape, lateral coherence, and other seismic attributes. If these changes in the seismic attributes can be identified and linked to the changes in sedimentary facies, the power references for sedimentary facies analysis are available.The traditional seismic facies analysis and seismic sequence analysis obtain sedimentary information by extracting geometrical relationship, contacting relation and attributive character of seismic event. This method which mainly depends on seismic facies reading works well on macro-scale. However, its usability is affected by the seismic resolution on a micro-scale (reservoir-scale). When we analyze the seismic facies on a reservoir-scale, reading seismic facies is not available. The reliable seismic facies analysis method (such as waveform clustering) can’t get a good result due to the excess noise caused by narrow time window (less than a half of phase). Because of the low seismic resolution and thin reservoirs of the research area, traditional seismic stratigraphy method is unable to meet the requirement of sedimentary microfacies study. So seismic sedimentology is introduced into this study.Seismic sedimentology is a developing subject whose research content is defined as "using seismic data to study sedimentary rocks and sedimentation" and often used in micro-scale (reservoir-scale) oil-gas exploration. Its core technology is still developing. Many seismic sedimentology methods obtain good effect in particular locations. But those methods usually can’t be applied widely due to their bad universality and tight applying conditions. Only two core technology are widely used after it has been introduced to china. There are few study on the applicability and validity of seismic sedimentology technology. For the thin reservoir of top Leikoupo Formation in research area and low frequency of seismic data, we focus on the seismic facies analysis in a narrow time-window used by waveform clustering, multiple attribute clustering, frequency division, EMD method and so on. We comparative study the applicability and effectiveness of those methods under different conditions by bulk testing. The waveform clustering method based on SOM and EMD is developed to provide precise reference for fine reservoir characterization. The detailed research contents and results in this paper are as follows:1. The single well sedimentary microfacies analysis and sequence division for Chuanke-1, Xiaoshen-1 and Xinshen-1 are completed by analyzing core, mud logging, well logging and sedimentary setting information. According to single well facies analysis, open platform is the main sedimentary facies in second half Member 3 of the Middle Triassic Leikoupo Formation. And restricted platform is the main the main sedimentary facies in first half which contains beach, tidal flat and lagoon. There are two third-order sequences in the Middle Triassic Leikoupo Formation. The Member 1 and 2 is the one, the Member 3 and 4 is the other one.2. Typical seismic facies are generalized by analyzing the seismic wave group feature near the well. We analyze the weathering crust at the top Leikoupo Formation and summarize its reservoir feature on well logging, rock physics and seismic facies. Low frequency, medium to high amplitude, chaotic seismic reflection and top lap are the typical seismic facies characteristics of Karst reservoir. These features can be identified by waveform clustering to reveal the reservoir distribution.3. The seismic events of the karst stratum at top Leikoupo Formation presents nice quality. There are a few seismic noise and interpretation noise produced by horizon picking on them. In this case, seismic facies analysis on a narrow window can be applied. The main frequency of seismic data of Leikoupo Formation in research area is about 26 Hz and the narrow time window is less than a half phase (about 19.2ms) conversely, there are some noises on the other part of Leikoupo Formation due the gyprock. We can only use seismic facies analysis with a wider window. Seismic facies analysis used by different methods is tested for different interval intvls and different time-windows. Then we draw following conclusions:1) Waveform clustering tends to have a stable and reliable result but is sensitive to seismic noise and interpretation noise when we use a narrow time-window (10-20ms). The displaying results of waveform clustering can’t be obtained until the width of the window reaches 20ms in research area. This width depends on the complexity and noise of the seismic events.2) Multiple attribute clustering is not sensitive to noise. So we can obtain a fine displaying results when use a 10ms time-window. The results of multiple attribute clustering differs from another and are unstable. If appropriate attributes can be extracted to serve as the input, the results of multiple attribute clustering would provide a reliable result.3) When we use the seismic data volume which contains high frequency component as the input for waveform clustering, the result shows more details and noisy. We can screen the frequency range which have a good result by using waveform clustering on each frequency. Then the seismic data volumes in this range can be reconstructed for waveform clustering to get better details and effect.4) Time window selection. When we do seismic facies analysis, no matter Waveform clustering or multiple attribute clustering, variable window thickness is a better choice than uniform window thickness. This window extract the information of target zones and avoid taking in the information of other zones. This greatly reduces noise and unreliability of interpretation.5) Attributes selection of multiple attribute clustering Instantaneous frequency, which have a stability in attribute clustering with a narrow window, is useful to identify the reservoir. Instantaneous phase has a high requirement on time-window in clustering. A bad choice on time-window may leads to a completely wrong result. Therefore, Instantaneous phase is better to be eliminated when we use uniform thickness window.4. For the sensibility of waveform clustering with a narrow time-window to seismic noise and interpretation noise, the waveform clustering method based on SOM and EMD is developed. We first decompose seismic data into IMFs using EMD method. Then, the IMFs which contain the fine information and smooth noise are chosen to reconstruct a new seismic data. Next, we put the reconstructed data as the input to SOM training and clustering. Our proposed method is less sensitive to noise especially whether in model test or in application to real seismic data from western Sichuan. It enhance the vertical resolution of seismic facies analysis and provide a better result for waveform cluster analysis with a narrow window (10ms). Therefore, this method can solve the problem that the waveform clustering has a bad result with a narrow time window to some extent and is a new technology of seismic sedimentology.5. With the method testing and analysis, waveform clustering method based on SOM and EMD is optimized for seismic facies analysis on top Leikoupo Formation karst reservoir with a 10ms time-window. The results of this method show the vertical variations of microfacies in two parts and offer reference to fine reservoir characterization. And multiple attribute clustering is chosen to accomplish the seismic facies analysis for Member 3 and 4 of the Middle Triassic Leikoupo Formation. Then, the results are interpreted to analyze the sedimentary evolution process cooperate with depositional setting of research area.