摘要
储集层地质建模是油藏描述中的一项基础研究，目前已发表的随机模拟方法已有十几种。我国陆相高度非均质储集层极其复杂，在进行地质建模之前，必须选择合适的随机模拟方法。以胜利油区孤岛油田中一区馆陶组上段４４小层河道砂储集层为研究层，以１３３口井开采初期的实测含油饱和度为原始数据，分别用指示克立格插值法、顺序指示法、马尔柯夫贝叶斯指示模拟法、序贯高斯模拟法、退火模拟法（原始数据）、退火模拟法（正态变换后数据）进行模拟，根据每一种方法１００次模拟结果绘制预测等值线图，比较６种模拟方法的模拟效果；再抽稀出８口井作为检验井，用上述６种方法重新模拟，对比各井的含油饱和度实际数据与各方法的预测值，用预测相对误差来检验各方法对河道砂储集层建模的适应性和有效性。认为在已有的条件数据控制下，各方法都能较合理地预测出参数分布的变化情况，但前３种指示模拟法更能反映局部的更小范围的非均质变化。
Reservoir geological modelling is a basic research in reservoir characterization. At present time, more than 10 kinds of stochastic simulation method have been developed. Because strongly heterogeneous continental reservoirs in China are very complicated, appropriate stochastic simulation methods must be chosen before geological modelling. Ng 4 4 channel sandstone reservoir in the upper part of Guantao Member, Middle 1 unit of Gudao oil field, Shengli oil field complex is used as a case study. Indicator Kriging, sequential indicator simulation, Markov Bayes indicator simulation, sequential Gaussian simulation, simulated annealing (original data) and simulated annealing (normal transformed data) are separately applied to geological models using actual oil saturation data from 133 wells in earlier oilfield development. According to the 100 simulated values for each method the predicted contour maps are first drawn, and then the simulation results of 6 methods are compared, finally 8 wells are selected and above 6 methods are used again to compare the actual oil saturation values with predicted values for 8 wells. The predicted relative errors are employed to check the effectiveness of channel sandstone reservoir modelling. This study shows that the distribution of reservoir parameters can be predicted reasonably using these methods under the existing data constrain conditions, however, local and small scale reservoir heterogeneity can be reflected only using the first three methods.
出处
《石油勘探与开发》
SCIE
EI
CAS
CSCD
北大核心
1998年第4期62-64,共3页
Petroleum Exploration and Development
关键词
储集层
随机模拟
分析
油藏
Channel sand, Reservoir, Oil reservoir description, Stochastic process, Analog, Algorithm analysis
