智能相控波形指示反演技术及其在煤系地层薄砂岩预测中的应用

Intelligent facies-controlled seismic meme inversion and its application in predicting thin reservoirs in coal measure strata

1.中石油煤层气有限责任公司,北京100028;
2.中联煤层气国家工程研究中心有限责任公司,北京100095;
3.北京中恒利华石油技术研究所,北京100102

1. PetroChina Coalbed Methane Company Limited,Beijing 100028,China;
2. Zhonglian Coalbed Methane National Engineering Research Center Co.,Ltd.,Beijing 100095,China;
3. PEGETE Group Inc.,Beijing 100102,China

针对煤系地层薄砂岩预测的难题,在精细分析地震波形与测井曲线关系的基础上,建立了基于地震波形横向相似性分析的智能相控波形指示反演技术。首先,按照相似性原理建立了不同的地震波形样本组,建立表征不同地震相类型的多个贝叶斯反演模型。其次,对每个样本组的测井曲线进行滤波,提取低频共性结构作为贝叶斯反演的先验信息,建立地震反演高分辨率初始模型。最后,通过不断修正初始模型,并利用实际地震数据进行验证,获得高分辨率的反演结果。正演模型验证结果显示,智能相控波形指示反演技术能够准确预测厚度为3m的煤系地层薄砂岩,证明了该反演方法的合理性和反演结果的准确性。鄂尔多斯盆地D地区煤系地层薄砂岩反演实例表明,智能相控波形指示反演技术对煤系地层薄砂岩预测精度达到了2~4m,后验井反演符合率达到86.96%,为煤系地层薄砂岩预测提供了一种全新的思路。

In response to the challenge of predicting thin reservoirs in coal measure strata,we analyze the relationship between seismic waveforms and log curves,and propose an intelligent facies-controlled seismic meme inversion technique based on the lateral similarity of seismic waveforms.According to the principle of similarity,the sample sets of different seismic waveforms are established,followed by a number of Bayesian inversion models representing different seismic facies types.A high-resolution initial model is established as the prior information of Bayesian inversion by performing lowpass filtering of log curves for each sample set.The initial model is iteratively refined using seismic data to obtain the final high-resolution inversion result.Forward modeling-based validation demonstrates that coal measure thin sandstones of 3 m thick could be predicted accurately using the intelligent inversion technique.A case study in the Ordos Basin shows that thin sandstones of 2~4m thick could be predicted and the success rate of prediction reaches 86.96% for those wells not involved in the inversion.This technique provides a new approach for thin sandstone prediction in coal measure strata.

中国石油集团“十四五”重大专项(2021DJ2002(JT))资助。

10.12431/issn.1000-1441.2025.64.01.010