Adding additives into the furnace is an effective way to reduce particulate matter (PM) formed during coal combustion. However, a numerical model that can predict the performance of additives on PM reduction is still lacking. In this work, a numerical framework, which contains the submodels for ash formation and ash–additive interactions, was developed to predict the ash particle size distribution with and without kaolin additives. The model fully considers the physical–chemical reaction process between the combustion gaseous products (alkali metal vapor) and the Si-/Al-based adsorbent particles. The simulation results are validated by experimental measurements of PM formed during the combustion of high-sodium Zhundong coal on a lab-scale furnace. Given the typical conditions of the lab-scale furnace, adding 3% (by weight) kaolin reduces the mass yield of PM0.3 by 33.35% but has no obvious influence on PM>1. On the basis of a time-scale analysis, we show that the chemical adsorption of the mineral precursors is the dominant mechanism for PM reduction. The influences of the kaolin particle size and dosage on PM reduction are evaluated. The PM reduction efficiency quickly increases and then enters into a plateau as the kaolin mass ratio increases. When a certain critical value is reached, a further increase of the addition ratio has a negligible effect on PM reduction. The critical kaolin mass ratio is 4.1% for 2.1 μm of kaolin and quadratically increases to 4.9% for 21.3 μm of kaolin. Increasing the kaolin size, in contrast, has a negative impact on PM reduction. On the basis of the simulation results, it can be concluded that there exists an optimal condition for PM1 reduction by kaolin addition. The numerical framework developed in the current work can help to quickly find the optimal addition strategy under different additive/coal properties, combustion temperatures, and atmospheres.