基于神经网络的涡发生器设计与试验研究

Design and Experimental Study of Vortex Generator Based on Neural Network

具有体积小、质量轻、分离效率高及能效比低的轴向涡流分离器的研究受到广泛关注,然而,至今尚无其涡发生器结构设计的文献报道。鉴于此,采用CFD数值模拟和BP神经网络相结合的方法分析了高分离效率下,涡发生器结构参数随关键操作工艺参数间的变化趋势,并依次确定了特定待分离混合液物性参数下最优结构参数组合。最优结构参数下轴向涡流分离器的分离效率随涡发生器转速先缓慢增加而后快速下降,最高分离效率可达80%以上,且CFD数值模拟结果和BP神经网络预测结果相吻合,最大误差仅有8.67%。利用最优涡发生器结构参数设计的轴向涡流分离器在中原油田进行了现场试验研究,试验结果与CFD数值模拟结果一致,两者间的平均误差仅为2.46%。综合理论分析和试验研究结果可知:轴向涡流分离器具有较好的分离性能,但其涡发生器的结构需根据待分离混合液物性参数和设备处理量进行针对性设计。

The research on the axial vortex separator with the features of small volume, light weight, high separation efficiency and low energy efficiency ratio has been paid extensive attention. H o w e v e r, there is no report on the structure design of the vortex generator. In view of this, the combination of C F D numerical simulation and B P neural network was used to analyze the trend of the structure parameters of the vortex generator with the key operat-ing parameters under high separation efficiency. The optimum combination of structural parameters has been deter-mined under the mixed liquid with specific physical property parameters. The separation efficiency of the axial vor- tex separator increases slowly and then decreases rapidly with the vortex generator rotation velocity,and the maxi- mum separation efficiency can reach above 80%. The C F D numerical simulation results are in agreement with the BP neural network prediction results, with the m a x i m u m error of only 8. 67%. The axial vortex separator with the optimal vortex generator structural parameters has been conducted field test in Zhongyuan Oilfield. The test results were in agreement with the CFD numerical simulation results, and the average error between them was only 2. 46%. The results show that the axial vortex separator has good separation performance, but the structure of the vortex gen-erator needs to be designed according to the physical properties of the mixture to be separated and the treatment ca-pacity of the equipment.