Solid particle erosion is a common and challenging phenomenon during the production and transport of particle-containing fluids and it is important to have models for predicting erosion rates accurately, especially for geometries such as elbows. Mechanistic models aim at predicting erosion accurately with low computational cost. In this study, a new particle trajectories-based mechanistic model is proposed to address the issues of liquid-dominated flows and the effect of particle size. Detailed flow and particle information for a standard elbow with water and air and different particle sizes are extracted from computational fluid dynamics (CFD) and analyzed to obtain a representative trajectory. The developed model includes various components that are sensitive to the particle size and flow conditions and accounts for the angle of impact and the turbulence in the flow. The proposed model is examined against CFD predictions for different pipe and particle sizes, and velocities with air, water, high-pressure air, and high-viscosity liquid. In comparison to an available mechanistic model, the new model provides relatively lower errors in predicting maximum erosion for many flow conditions. Moreover, the proposed model is found to be more consistent with CFD data for high-pressure air and higher-viscosity liquids. The model is further validated with experimental data for various conditions. Comparisons against numerical and experimental data suggest that the proposed model provides a significant improvement for liquid–solid flows and small particles.