摘要
影响压裂井的产能因素较多,采用常规的线性方法进行的产能预测效果不佳。由于研究区无压裂参数,鉴于神经网络技术高度复杂的非线性动力学系统功能,因此采用神经网络技术进行产能预测。优选有效孔隙度、有效渗透率、可动油饱和度、有效厚度、原油黏度、产水率以及初期日产油7个参数作为模型的训练样本,利用该模型进行预测。实践结果表明,该方法预测产能与试油日产油量符合率达到90%,很好地实现了在无压裂参数条件下,对低孔低渗砂岩储层压裂井产能的预测。
There were many factors affecting the productivity of fractured wells,and the effect of productivity forecast using the conventional linear methodology was no good.For there was lacking of fracturing parameters in the studied area and the highly complex functions of non-linear system in the neural network technology,therefore the neural network technology was used to forecast productivity in the studied area.Seven parameters as effective porosity,effective permeability,movable oil saturation,effective thickness,crude oil viscosity,water production rate and tested productivity of single layer were selected as training samples for the model and it was applied for productivity forecast.Practical results show that the forecasted productivity using the method is consistent with the daily oil output of production testing with the coincidence rate of 90%.The neural network technology is successfully applied for productivity forecasting without fracturing parameters.
出处
《石油天然气学报》
CAS
CSCD
2014年第4期107-110,共4页
Journal of Oil and Gas Technology
关键词
低孔低渗
砂岩储层
压裂井产能
BP神经网络
旬邑-黄陵地区
low porosity and low permeability
sandstone reservoir
productivity forecast
BP Neural Network
XunyiHuangling