Liquid–liquid phase separation by gravity is governed by complex local coalescence mechanisms often oversimplified in conventional design. Traditional models frequently assume ideal plug flow and "coalescence-by-contact," presupposing that any droplet impinging on an internal surface is immediately separated. This approach ignores the critical impact of local hydrodynamics and coalescence phenomena, which dictate the actual interaction and trajectory of droplets within the internal geometry. This work bridges the gap between theoretical sedimentation and real-world performance by analyzing surface-induced coalescence in angled plate packs through a combined experimental and numerical approach.

The experimental campaign utilized a DN150 separator with 3, 10, and 20 angled profiles at spacings of 7, 15, and 25 mm. Two water–oil systems (11 mPa·s and
200 mPa·s) were tested at a constant 1.0 vol.-% concentration. In-line photo-optical shadowgraphy (SOPAT) was employed to capture real-time droplet size distributions (DSD) at flow rates from 130 to 800 L/h.

Results show that while [...] improves with decreased plate spacing, increased pack length, and lower velocities, performance remains consistently below ideal predictions. Observations reveal that droplet impingement does not guarantee coalescence; instead [...]. These findings were used to develop a specialized coalescence model within a CFD framework. By accounting for local hydrodynamics rather than ideal flow conditions, the model demonstrates significantly improved predictive accuracy and shows good agreement with experimental data, providing a robust framework for the precise scale-up of industrial liquid-liquid phase separators...