Regression models for assessing the efficiency of vibratory separation of parsnip seeds taking into account air dynamics based on numerical simulation and field experiment

Abstract

To reduce the complexity of research into designing promising vibratory machines while minimizing the harmful effect of the aerodynamic factor, it is convenient to use regression models. With their help, a quantitative assessment of the effectiveness of separation (cleaning) of seed mixtures is carried out, depending on the design parameters and the mode of operation of vibratory machines. This paper reports the results of research on the construction of regression models for parsnip seeds based on numerical modeling and full-scale experiment. Based on numerical modeling, a four-factor regression model of the second order was built, which takes into account the geometric characteristics of the aerodynamic screen, the design of the set of working surfaces, and the oscillation amplitude of a vibratory machine. Based on a full-scale experiment, a three-factor regression model of the second order was constructed for a constant gap between the working surfaces. A comparative analysis of the resulting regression models suggests that numerical modeling provides satisfactory accuracy in assessing the influence of the aerodynamic factor. This estimate, when using a regression model based on a numerical experiment, exaggerates the estimate determined by the full-scale experiment by 5–15 % (depending on the regressate variation area localization). Hence, the numerical model of the process of vibrational motion of light-weight seeds, taking into account the action of aerodynamic forces and moments, used to build a regression model of separation of parsnip seeds, can be considered adequate. Regression models (for parsnips and other plant crops), which are built on the basis of numerical modeling, should be used to solve problems of optimizing the parameters of vibratory machines according to the criterion of reducing the harmful effect of the aerodynamic factor.