The study presents a comprehensive direct numerical simulation of gas‑liquid capillary flow in a coiled flow inverter (CFI) geometry, combining the effects of Dean and Taylor flows. Using a finite‑element method, the authors resolved the full three‑dimensional velocity field and tracked passive particles seeded at defined starting positions to quantify mixing and dispersion. The CFI model consists of a helical capillary that begins with a 360° turn, incorporates a 90° bend, and continues with a second 360° turn. Inner diameters and coil diameters were varied, while the liquid (water) and gas (air) properties were set to ρₗ = 1000 kg m⁻³, ρ_g = 1.2 kg m⁻³, μₗ = 1.12 mPa s, μ_g = 0.018 mPa s, and a reduced surface tension σ = 2.5 mN m⁻¹ to maintain numerical stability. Flow velocities of 3, 6, and 12 cm s⁻¹ were investigated, corresponding to Reynolds numbers Re = 42.86, 85.71, and 171.43. The phase ratio of liquid to gas was approximately 2.4:1, with bubble lengths of about 2.5 mm and slug lengths of roughly 6 mm. The simulation revealed that at these low Reynolds numbers the secondary Dean vortices, driven by centrifugal forces in the curved sections, dominate the mixing process. A high flow velocity already after two coils and the 90° bend produced a nearly complete and homogeneous distribution of the dispersed phase, whereas the coil diameter had a secondary influence, mainly through its effect on the dimensionless Dean number. Particle dispersion histograms showed that the dispersion induced by the 90° bend is more strongly affected by the flow velocity than by the coil diameter. The authors introduced a transformation method that constructs a periodic fluid domain and facilitates post‑processing of mixing characteristics, proving robust across a wide range of flow parameters. The study concludes that the developed multiphase flow model accurately captures the coupled Dean and Taylor phenomena in CFI geometries and that mixing efficiency can be significantly enhanced by increasing the flow velocity, with the coil diameter playing a lesser role.

The project was carried out at the Technical University of Dortmund, with computational work performed on the LiDO cluster and data management supported by the RDM team at TUDO. The research team comprised OM, RM, and ST, who led the numerical simulations and analysis, and JS and NK, who contributed experimental work and validation of the numerical results. Funding was provided by the German Research Foundation (DFG) under the priority program SPP 1740 (grant TU 102/53‑1) for the computational aspects and by DFG grant KO 2349/13‑1 for the experimental component. An additional DFG DEAL contract supported the open‑access publication. The simulation data have been deposited in the TU Dortmund dataverse (DOI 10.17877/tudodata‑2024‑lwoxtgmw), ensuring reproducibility and accessibility for future studies. The collaboration demonstrates a successful integration of advanced CFD techniques, experimental validation, and data stewardship within a funded research framework.