A skew normal-Laplace model of partition curve based on probability characteristics.

• Probability characteristics of the partition curve were analyzed. • The differential partition curve is asymmetric and fat-tailed. • A skew normal-Laplace model with three parameters was developed. • Accuracy and ability of this model were validated by comparison with other models. • The parameters transformation law when the separation density changed was provided. The mathematical model of partition curve is fundamental to evaluate and predict the gravity separation performance in coal cleaning industry. Existing models have achieved good fitting accuracy but lack explicit physical meanings. From the perspective of probability and error distribution, the common characteristics of actual partition curve and its differential curve were analyzed according to the generalized central limit theorem. A skew normal-Laplace (SNL) model with three parameters was established based on asymmetry and fat tail characteristics. Parameters in the SNL model represent location, scale and skewness of partition curve, which reflect separation density and error distribution, therefore it is convenient to evaluate the efficiency of the whole process. Compared with the existing models, the SNL model achieved smaller deviation with fewer parameters and conformed better to physical reality. For practical application, the parameters transformation law when separation density changed was provided according to their properties. [ABSTRACT FROM AUTHOR]