Liquid-spray cleaning has recently been considered an eco-friendly cleaning method in the semiconductor industry because it efficiently cleans contaminated wafers without using any chemicals, relying instead on direct momentum transfer through dropwise impaction. Previous researches are mainly divided into two groups, such as modelling studies predicting the cleaning effect of single-droplet impact and experimental works for measuring particle removal efficiency (PRE) that essentially accompanies multiple droplet impacts. Here, we developed a Monte Carlo model to connect the single-droplet based model to the ensemble effect of multiple droplet impacts in real cleaning experiments, and thereby predict the PREs from the impaction conditions of droplets and the diameters of target particles. Additionally, we developed a two-fluid supersonic nozzle system, capable of spraying 10-60 ㎛ droplets under control of impact velocity, with aims to validate the model predictions of PREs for 15-130 nm contaminant particles on a Si wafer. We confirmed that the model predictions are in agreement with the experimental data within 7% and the cleaning time needs to be controlled for ensuring the efficient removal of particles.