Numerical simulation of soot formation in a laminar premixed burner-stabilised benchmark ethylene stagnation flame was performed with a new detailed population balance model employing a two-step simulation methodology. In this model, soot particles are represented as aggregates composed of overlapping primary particles, where each primary particle is composed of a number of polycyclic aromatic hydrocarbons (PAHs). Coordinates of primary particles are tracked, which enables simulation on particle morphology and provides more quantities that are directly comparable to experimental observations. Parametric sensitivity study on the computed particle size distributions (PSDs) shows that the rate of production of pyrene and the collision efficiency have the most significant effect on the computed PSDs. Sensitivity of aggregate morphology to the sintering rate is studied by analysing the simulated primary particle size distributions (PPSDs) and transmission electron microscopy (TEM) images. The capability of the new model to predict PSDs in a premixed stagnation flame is investigated. Excellent agreement between the computed and measured PSDs is obtained for large burner-stagnation plate separation (≥ 0.7 cm) and for particles with mobility diameter larger than 6 nm, demonstrating the ability of this new model to describe the coagulation process of aggregate particles.