论文详情
储层含油气后, 地震数据的振幅、频率等信息会发生变化, 这为利用地震数据进行油气识别奠定了基础。根据互补集合经验模态分解(CEEMD)算法能够自适应分离出复杂信号的局部特征信息的优势, 提出利用该算法提取地震数据中的烃类信息。分别采用经验模态分解(EMD)算法、集合经验模态分解(EEMD)算法以及CEEMD算法对MarmousiⅡ模型合成记录及实际测井数据合成记录进行分解, 发现: 基于EMD算法得到的各阶固有模态函数(IMF)分量与油气无明显对应关系; 而在基于EEMD算法和CEEMD算法得到的IMF分量中, IMF1分量能够突出油气层的地震响应, 压制非油气层处的地震响应, 但在集成次数较少时, EEMD算法对应的IMF1分量受白噪声影响较为明显。相比EMD算法、EEMD算法, 在较少计算量下, 基于CEEMD算法得到的IMF1分量能够高精度刻画烃类的分布范围。最后将该方法应用于渤海A油田, 根据该方法得到的烃类识别结果, 部署A2井落实油水界面, 该井实钻0.9m油层、1.3m油水同层及6.1m水层, 钻前预测结果与实钻结果基本一致, 证实了方法的可行性与有效性。
When reservoirs contain oil and gas, the amplitude and frequency of seismic data change, thereby laying a foundation for oil and gas identification through seismic data. Because the CEEMD algorithm can adaptively separate the local feature information of a complex signal, this study adopted this algorithm to extract hydrocarbon information from seismic data. By adopting EMD, EEMD, and CEEMD algorithms to decompose Marmousi Ⅱ model synthetic records with different dominant frequencies and phases and actual logging data synthetic records, it was revealed that there is no obvious correspondence between the IMF components based on EMD and oil and gas. Among the IMF components obtained based on EEMD and CEEMD, the IMF1 component can highlight the seismic response of oil and gas reservoirs and suppress the seismic response of non-oil and gas reservoirs. However, when the number of integrations is small, the MF1 component based on EEMD is significantly affected by white noise. Compared with EMD and EEMD, the IMF1 component based on the CEEMD algorithm can accurately depict the distribution range of hydrocarbons with fewer computations. Finally, the method was applied to the Bohai A oilfield, and the results showed that A2 well is deployed to implement oil-water contact. The A2 well has a 0.9m drilled oil layer, 1.3m oil-water layer, and 6.1m water layer. The pre-drilling prediction results were consistent with the actual drilling results, which confirmed the feasibility and effectiveness of the method proposed in this study.