数值模拟历史拟合是数值模拟研究中非常重要的一部分,是不同开发方案或调整方案对比优选、油气产量预测的基础。目前历史拟合工作主要采用传统的人工反复试算方法,既复杂又费时费力,存在多解性。为了提高数值模拟拟合精度及工作效率,通过对计算机辅助历史拟合技术方法进行研究,以Eclipse为模拟器,以集合卡尔曼滤波方法为技术基础,从模型输入数据文件自动获取数据,将敏感性参数作为被估参数,确立目标函数,利用最优化方法实现自动修正可调参数,自主研发了数值模拟自动辅助历史拟合软件。实例应用表明,通过自主研发的辅助历史拟合软件实现了多模型自动历史拟合,采用最优化算法筛选出最优拟合模型,节省了大量人工拟合时间,提高了拟合精度及工作效率。提出的数值模拟辅助历史拟合方法可行可靠,取得了良好的应用效果。

The historicalmatchingof numerical simulation is a very important part of numerical simulation research, which is the basis of comparison and optimization of different development schemes or adjustment schemes and prediction of oil and gas production.At present, the historicalmatchingmainly adopts the traditional manual trialcalculation, which is complex, time consuming and labor consuming, and has multiple solutions.In order to improve the precision and efficiency of numerical simulationmatching, through the study of computer-aided historymatchingtechnology,Eclipseis selectedas a simulatorwith ensembleKalman filter method set as the technical basis, automatic data acquisitionis achievedfrom the model input data file,whilesensitivity parametersare setas estimated parameterstoestablish the objective function,andto achieve automatic correction of adjustable parametersbyusing the optimization method.The numerical simulation aided historical matching software has been developed independently.The application shows that the multi-model automatic historymatchingis realized by the self-developed historymatchingsoftware, and the optimalmatchingmodel is selected by using the optimizedalgorithm, so that a large amount of manualmatchingtime is saved, and the precision and working efficiency are improved. The numerical simulation aided history fitting method is feasible and reliable, and has achieved good application effect.