Developing a dimensionless number for solid particle erosion with gas-liquid-solid flow in standard elbows

Abstract

A new dimensionless number for erosion prediction in standard elbows under gas-liquid-solid flow conditions is developed. For this purpose, a machine learning algorithm is trained and used to generate erosion data. Then, three important dimensionless groups are extracted employing dimensional analysis. Next, self-similarity is established among these groups and a new dimensionless number (PnrTP) is developed. It is revealed that erosion can be predicted through mixture Reynolds number for different flow regimes, when the deviation from the homogenous no-slip conditions is accounted for. Furthermore, validation is performed against different experimental data for a wide range of flow and particle conditions.