An analysis method of injection and production dynamic transient flow in a gas field storage facility

A dynamic transient flow analysis method considering complex factors such as the cyclic injection and production history in a gas field storage facility was established in view of the limitations of the existing methods for transient flow analysis and the characteristics of the injection-production operation of strongly heterogeneous gas reservoirs, and the corresponding theoretical charts were drawn. In addition, an injection-production dynamic transient flow analysis model named "three points and two stages" suitable for an underground gas storage(UGS) well with alternate working conditions was proposed. The "three points" refer to three time points during cyclic injection and production, namely, the starting point of gas injection for UGS construction, the beginning and ending points of the injection-production analysis stage;and the "two stages" refer to historical flow stage and injection-production analysis stage. The study shows that the dimensionless pseudo-pressure and dimensionless pseudo-pressure integral curves of UGS well flex downward in the early stage of the injection and production process, and the dimensionless pseudo-pressure integral derivative curve is convex during the gas production period and concave during the gas injection period, and the curves under different flow histories have atypical features. The new method present in this paper can analyze transient flow of UGS accurately. The application of this method to typical wells in Hutubi gas storage shows that the new method can fit the pressure history accurately, and obtain reliable parameters and results.

基于改进联合分布适应的轴承智能故障诊断方法

在轴承故障诊断过程中,存在缺乏足量故障样本、变工况下信号分布差异等问题。虽然基于机器学习和深度学习方法的智能故障诊断方法的运用取得了许多成果,但该方法在应用过程中仍面临一些挑战,阻碍了智能故障诊断方法在实际工业场景下的应用。为此,提出了一种基于改进联合分布适应的轴承智能故障诊断方法(BIFD-IJDA)。首先,利用小波包变换对振动信号进行了分解与重构,再计算了重构信号的统计参数,构成了原始特征集;然后,设计了基于特征重要度与KL散度的迁移特征选取方法,对各统计参数特征进行了量化评估;采用了改进联合分布适应方法,对源域和目标域特征集进行了分布适应处理,降低了域间分布差异;最后,利用源域特征样本训练的故障诊断模型预测了目标域样本故障类别,采用美国凯斯西储大学实验台和机械故障模拟(MFS)实验台的轴承故障数据,开展了不同工况下的故障诊断实验。实验结果表明:该故障诊断方法在2种轴承故障数据下取得的最大故障诊断准确率分别为100%和96.29%,明显优于其他对比模型。研究结果表明:该故障诊断方法具有应用于实际工业场景的潜力。

大庆油田气井排水采气技术研究进展及下步攻关方向

通过总结大庆油田排水采气技术领域“十三五”期间重要研究进展,分析当前面临的老区稳产维护工作量大、新区含硫井排采难、气田地面水处理压力大等诸多排水采气难题,提出下步重点攻关方向。“十三五”期间,针对火山岩气藏底水活跃、气井见水快导致产量大幅下降和S气田单井产量低、积液井数多、长期低效开发的问题,攻关形成了排水采气技术系列,实现了气井由“零散被动、单井排水”到“积极主动、整体治水”的跨越。“十四五”期间,在老井稳产综合治理技术、含硫气井排水采气及解堵技术、气田水地面处理技术方面仍面临着诸多挑战。针对技术问题,提出气井工况智能分析与远程测控技术、高压高含硫气井生产工艺技术、气井井下气液分离同井注采技术3个重点研究方向,为大庆油田气井效益开发提供工程技术支撑。