应用随机反演改善储层表征的过程(英文)

The application of stochastic inversion to improving the reservoir characterization process

为了适当地完成储层表征的过程,一个有效的方法就是把现场所有可以利用的信息融合成一个一致性的模型。在实际生产中实现这种融合并非简单的任务,所以有必要运用如地震反演等特殊方法。应用地震反演可以使测井数据和地震数据的有效结合成为可能,并且可以得到一个模型,该模型在预测过程中可通过流体数字模拟来验证。地震反演可以通过多种方法进行,主要分为两大类:一类是确定性方法(其代表是回归反演和约束稀疏脉冲反演),另一类是随机方法(其代表是地质统计学反演)。在本次研究中,通过随机反演结果和确定性反演结果的对比展示了随机反演是如何改进储层表征过程的。事实上,随机反演,可以运用较高的采样率(和储层模型的网格大小相接近),来产生一个更可靠的模型。随机反演的另一个好处就是随机方法可产生一些基本的统计测量值来改进解释精度,并且在储层表征过程中能生成大量的实现,从而使储层模型的不确定性研究成为可能。

To properly accomplish the reservoir characterization process in an effective way, it is necessary to integrate all the available information about the field in a consistent model. It is not an easy task to perform this integration in practice： it is necessary to use some specific methods such as seismic inversion. The seismic inversion is an efficient method to integrate both well-log and seismic data, obtaining a model that could be used in a forecasting process by numeric flow simulation. The seismic inversion can be achieved by several methods, which can be divided into two main groups： the deterministic methods （represented by the reeursive inversion and constrained sparse-spike inversion） and the stochastic methods （represented by the geostatistical inversion）. In this study, we show how stochastic inversion can improve the reservoir characterization process, by comparing its results with those obtained by deterministic inversion. The stochastic inversion can use a high sample rate that is close to the cell size of the reservoir model, implying that a more reliable model can be generated. The stochastic inversion method can also generate some basic statistics measurements that improve interpretations. Due to the great number of realizations generated during the process, we could also perform an uncertainty study on the reservoir model.