Search

In the pump-free lithium bromide absorption refrigeration system, the two-phase flow pattern in the lift tube of the bubble pump has a decisive influence on the lift performance of the bubble pump. The heat and mass transfer between two phase interfaces can be accurately described by mesoscale method,In order to explore the working mechanism of lithium bromide solution bubble pump, the improved lattice Boltzmann method combine lattice Boltzmann heat transfer model are adopted to simulate The rising process and coalescence behavior of multiple bubbles in lithium bromide solution in a vertical bubble pump with gas-liquid density ratio of 2778. The density field and velocity vector distribution of multiple bubbles under different initial conditions were obtained. The critical coalescence distance of water vapor in lithium bromide solution with double bubbles and 3 bubbles is obtained. The motion process of 4 bubbles with different initial settings is simulated. The influence of the number of bubbles on the critical coalescence distance is discussed. That results provide a great reference value for the study of the transition mechanism between flow patterns.

In order to study the symbiotic medium of two petroleum degrading bacteria, Cycloclasticus sp. and Alcanivorax sp. Single factor experiment, uniform design and neural network combined with genetic algorithm were used to optimize medium components. The concentration of carbon source (diesel) is 0.5%, the concentration of nitrogen source (NH4)2SO4 is 8 g/L, phosphorus source is KH2PO4:Na2HPO4 mol ratio 3:1, yeast powder concentration is 0.03 g/L, culture time is 3D, culture temperature is 20, pH = 7.6, loading amount is 30 mL, inoculation amount is 1.5%, and rocking bed speed is 150 rpm. The test results were consistent with expectations. The degradation rate reached 89.80% after optimization and 14.4% higher than the original 78.50%. Through the optimization of the symbiotic medium, the culture medium suitable for the symbiosis of bacteria group is obtained, which shows that the neural network and genetic algorithm have remarkable superiority in the optimization of culture medium.

Oil spill identification is important for marine environmental protection. In this paper, the artificial neural networks (ANNs) model is built to analysis FTIR fingerprint for identifying oil spill sources. The BP neural network model was trained with the specific data from literature, and this kind of spill oil ANNs was used to predicate the unknown samples successfully. The identification accuracy rate of the ANNs was 96.5%. This method can be used to analyze the infrared fingerprint of oil and identify the source of oil spill at sea. When predicting 12 unknown samples, 67, 69 and 71 samples deviate, but the accuracy of prediction is not affected. The different training methods have great influence on the network output. According to the results, the ratio conjugate gradient algorithm and the Levenberg-Marquardt algorithm training network model have the highest accuracy, the ANNs can be used for various different species as well as the weathered oil samples.