【摘要】 现有储层建模过程中,往往是通过井点数据进行基于地质统计学的空间插值与模拟,即从数学认识到地质认识,这种仅仅单一依靠变量空间相关性的数学统计,难以将地质学家的认识反映到最终的地质模型当中。针对这一问题,提出了相控随机建模中利用虚拟井提高地质约束的方法。首先是在对区域整体地质认识的基础上,设定反映全区非均质特征的变差函数模型,利用克里金算法分析估计方差的分布特征,挑选估计方差分布较大的区域作为插入虚拟井的备选区,同时考虑局部的地质认识确定具体的虚拟井位置。其次利用波阻抗地震属性与孔隙度的相关关系给虚拟井赋值,然后再将虚拟井作为"硬数据"再次利用给定的变差函数模型进行克里金插值,再次挑选方差分布大的区域并设置相应的虚拟井。依次迭代以上步骤直到全区克里金估计方差分布平稳且均匀。最后利用已知井点和虚拟井点数据拟合变差函数进行属性建模,这样就将地质分析所得到的关于研究区的非均质特征体现在了后期的储层属性的建模当中。应用上述方法,以渤南盆地B油田Es4-23小层进行了相控储层的属性随机建模,实例结果表明,该方法能够有效提高模型精度,得到的模型更符合地质实际。最后对模拟结果的准确性进行了分析,讨论了造成模拟误差的原因及降低误差的方法。更多还原

【Abstract】 The common practice in reservoir modeling is to infer variables’ correlation from well data at first,and then to do spatial interpolation and stochastic simulation based on various Geo-statistical methods,which is the process from mathematical statistics to geological understanding.However,geological concepts can not be fully exhibited in the final reservoir model by merely using this kind of variable correlation.A new method is proposed in this paper to enhance the geological constraint by inserting virtual wells during stochastic reservoir modeling.Firstly,set the variogram model that can characterize the heterogeneity characteristics of the whole region based on the comprehensive study of geological background.And then,the distributive characteristics of estimation variance by using Kriging interpolation algorithm are analyzed.After that,larger value areas are selected to interpolate virtual wells.Meanwhile,the geological information is considered to pick the exact locations of virtual wells.Secondly,the property values(mainly porosity and permeability)of the virtual wells are given based on the seismic property of impedance-porosity.Thirdly,the Kriging interpolation is done again using the virtual well as"hard data",and then select relative larger estimation variance value areas and insert virtual wells again.Repeat the above two steps iteratively until the estimation variance distribution of the whole region is uniformly distributed.Finally,fit the variogram parameters based on virtual and existed well data together and then proceed property modeling.In this way,the former geological concepts can be smoothly merged into the process of reservoir modeling.Taking Es4-23 of B oil-field of Bonan Basin as an example,stochastic reservoir modeling is proceeded by adopting the method illustrated above.The result shows that this method enhances reservoir model’s accuracy effectively,which is more geologically reasonable.In the end,the accuracy of the outcome model is discussed,and modeling error’s causing reasons and the methods of diminishing error are analyzed.更多还原