An integrated model of the PUNQ-S3 reservoir in the North Sea was constructed. Then, 16 parameters with the highest impact on target functions were selected for sensitivity analysis. The sensitivity analysis was based on the two-level Plackett–Burman design of experimental method. Finally, seven variables with the highest impact on the target functions were selected. Net present value, cumulative oil production, and cumulative water production were three target functions. The proxy model was constructed using the three-level Box–Behnken experimental design for each of the three target functions, taking into account the effect of the variable’s interactions on each other. Then the compliance and predictivity of the proxy model for each target function were validated according to the decision variables. In the end, multiobjective optimization was conducted with the aim of maximizing net present value and cumulative oil production and minimizing cumulative water production using a parameter called composite desirability.

中文翻译：

---

基于使用DOE方法构建代理模型的油田综合生产优化

建立了北海PUNQ-S3水库的综合模型。然后，选择对目标功能影响最大的16个参数进行敏感性分析。敏感性分析基于实验方法的两级Plackett-Burman设计。最后，选择了对目标功能影响最大的七个变量。净现值，累计产油量和累计产水量是三个目标函数。使用三个级别的Box–Behnken实验设计针对三个目标函数中的每一个构建了代理模型，同时考虑了变量之间相互作用的影响。然后根据决策变量验证每个目标功能的代理模型的依从性和可预测性。到底，