Study on a check valve for airducts with a nonmobile guide vane based on a random forest model

The check valve is an indispensable local component in pipe network systems, but its active components are worn during use, increasing the energy consumption of the system and endangering the safety of the fan. The traditional Tesla valve structure poorly matches the pipeline in engineering practice. This paper proposes a check valve for airducts with a nonmobile guide vane that makes the pressure drops generated when the fluid flows through the valve from the forward and reverse directions significantly different. In this study, the diodicity mechanism was analysed, the valve shape was optimized by means of machine learning, numerical simulation and full-scale test verification, and the resistance coefficient ratio (ζ) was proposed to evaluate the check effect. A comparison was made with the diodicity of the Tesla valve. The results show that the check performance of the novel check valve proposed in this study is significantly better than that of the traditional Tesla valve, and its diodicity is 255% higher. When the fluid reaches full development, the resistance coefficient ratio reaches 4.93. The novel check valve is more suitable than the existing valve designs for the pipelines in engineering practice and provides theoretical support and necessary equipment for flow control in ventilation and air conditioning systems. •A check valve with a nonmobile guide vane is investigated by machine learning and numerical simulation.•The check effect is verified by the full-scale experiments.•The resistance coefficient ratio is proposed.•Compared with the traditional Tesla valve, the diodicity of the novel check valve increased by 255%.