Heuristic prediction of gas precipitation performance of self-excited oscillation cavity

Abstract

The precipitation of dissolved gas in oil is a challenging problem in pollution control of hydraulic systems. When the self-excited oscillation jet is formed, there are two low-pressure regions in the self-excited oscillation cavity, and the reduction in pressure causes the dissolved gas in the oil to precipitate out. Here, we investigated the effect of the self-excited oscillation cavity on the dissolution of dissolved gas in oil. We studied the gas precipitation performance of the self-excited oscillation cavity by simulating the pressure and velocity fields inside the cavity under different ratios of dimensionless structure parameters. The results indicated that parameter intervals for maintaining good gas precipitation performance of the self-excited oscillation cavity were d2/d1=2–2.4, D/d2=4–6, and D/L = 2. We then used a heuristic prediction algorithm (Genetic algorithm-backpropagation, GA-BP) to fit the simulation and experimental data, in which the root mean square error between the simulation and experimental data was only 2.45%. This indicated that the simulation of the flow field was reasonable, and that the GA-BP model performed well in predicting the gas precipitation performance of the self-excited oscillation cavity. Our results have important guiding significance for future studies on the gas precipitation performance of the self-excited oscillation cavity.