Machine learning approach to improve vapor recovery: Prediction and frequency converter with a new vapor recovery system

In practice, the volatile organic compounds (VOCs) pollution can exist when refueling due to the properties of the gasoline, low viscosity and high saturated-vapor pressure. A new gasoline vapor recovery system involving frequency conversion technology and machine learning is developed to cope with this problem. In the proposed system, firstly, the pumping capacity of the vacuum pump is evaluated, and test shows an almost linear relationship between suction volume and frequency. Then, the Multi-Layer Perception (MLP) neural network and the support vector regression (SVR) are employed to predict the gas-liquid ratio, and the numerical examples are presented to prove the high prediction accuracy of the MLP and SVR, respectively, where the MLP neural network has better generalization ability. Finally, compared with the two gasoline vapor recovery systems based on the 1: 1 fixed control model and the PID control model, respectively, the gasoline vapor recovery efficiency is improved significantly by the new gasoline vapor recovery system.

Prediction of Secondary Organic Aerosol from the Multiphase Reaction of Gasoline Vapor by Using Volatility–Reactivity Base Lumping

The secondary organic aerosol (SOA) formation from photooxidation of gasoline vapor was simulated by using the UNIfied Partitioning Aerosol phase Reaction (UNIPAR) model, which predicted SOA growth via multiphase reactions of hydrocarbons. The Carbon Bond 6 (CB6r3) mechanism was incorporated with the SOA model to estimate the hydrocarbon consumption and the concentration of radicals (i.e., RO2 and HO2), which were closely related to atmospheric aging of gas products. Oxygenated products were lumped according to their volatilities and reactivity and linked to stoichiometric coefficients and their physicochemical parameters, which were dynamically constructed at different NOx levels and degrees of gas aging. To assess the gasoline SOA potential in ambient air, model parameters were corrected for gas–wall partitioning (GWP), which was predicted by a qualitative structure activity relationship for explicit products. The simulated gasoline SOA mass was evaluated against observed data obtained in the UF-APHOR chamber under ambient sunlight. The influence of environmental conditions on gasoline SOA was characterized under varying NOx levels, aerosol acidity, humidity, temperature, and concentrations of aqueous salts and gasoline vapor. Both the measured and simulated gasoline SOA formation was sensitive to seeded conditions (acidity and hygroscopicity) and NOx levels. A considerable difference in SOA mass appeared before and after efflorescence relative humidity in the presence of salted aqueous solution. SOA growth in the presence of aqueous reactions was more impacted by temperature than that in absence of seed. The impact of GWP on SOA formation was generally significant, and it appeared to be higher in the absence of wet salts. We conclude that the SOA model in the corpus with both heterogeneous reactions and the model parameters corrected for GWP is essential to accurately predict SOA mass in ambient air.