Multiphase CFD case study
A phase-aware OpenFOAM study of how a rectangular-slot liquid–gas interface deforms under aerodynamic forcing across an early-time VOF window.
The contribution is a reproducible bridge from explicit CPU/MPI numerics to native-time phase fields, mixture-flow diagnostics, and topology and morphology measures across all 25 recorded states.
Research question and numerical setup
The calculation follows that response through native-time phase fields, mixture-flow diagnostics, cross-sections, and connected morphology while keeping the interpretation tied to the available early-time window.
The scaled high-density-gas case uses a gas-density factor of 20, a 12 m/s liquid inlet speed, and an estimated gas-side Weber number We_g = 53.3333. These parameters increase interface forcing while keeping the interpretation tied to this numerical configuration.
Flow proceeds in +z; x and yare transverse. alpha.water is liquid fraction, 1-alpha.water is its derived two-phase gas complement, and U is mixture velocity.
| Domain | Internal duct z = −15 to 0 mm; external domain 12 × 8 × 60 mm |
|---|---|
| Rectangular slot | 2 × 1 mm at z = 0; 16 × 12 cells across the slot |
| Mesh | 661,248 documented hexahedral cells; 684,983 VTK points |
| Solver | OpenFOAM v2406 interFoam; incompressible two-phase VOF |
| Model | Laminar configuration with alpha.water as liquid volume fraction |
| Time control | Adaptive step, max Δt = 2.083 × 10⁻⁶ s; max Co and alpha-Co = 0.3 |
| Numerics | Euler time integration; linearUpwind momentum; van Leer alpha advection; MULES and PIMPLE |
| Execution | Linux workstation, CPU/MPI, four subdomains; no GPU solver path |
| Analysis | Python, VTK, Matplotlib, ffmpeg, and headless Blender |
| Native output | 25 states at 0.0005 s intervals from 0 to 0.012 s |
Time evolution
The evolution contains 25 discrete native states from 0 to 0.012 s. Each state is held for legibility; no temporal geometry interpolation is used.
0 s0.0005 s0.001 s0.0015 s0.002 s0.0025 s0.003 s0.0035 s0.004 s0.0045 s0.005 s0.0055 s0.006 s0.0065 s0.007 s0.0075 s0.008 s0.0085 s0.009 s0.0095 s0.01 s0.0105 s0.011 s0.0115 s0.012 salpha.water=0.5 states, presented as labeled discrete holds at 24 fps with no temporal geometry interpolation.
Phase and flow diagnostics
Liquid fraction, its derived gas complement, and mixture velocity are analyzed separately so each figure retains its field meaning, threshold, units, and native time.
At the native 0.006 s state, connected curves are seeded deterministically in gas-rich (alpha.water ≤ 0.1) and liquid-rich (alpha.water ≥ 0.9) regions. They are instantaneous mixture-U curves conditioned on seed location.
Morphology and representative results
Surface metrics track area proxy, axial extent, component count, and largest-component dominance at every native state. The largest component retains a minimum face fraction of about 0.979 while the interface deforms over the early-time sequence.
alpha.water=0.5 face mesh at the exact native 0.0055 s state. Smooth motion comes from the camera path; the reconstruction adds no CFD resolution or physics.
Related work
Read-only native fields move through phase-aware checks, descriptive analysis, and deterministic figure or scene plans. Those requirements guide the adjacent CLI without changing the CFD case or its scientific interpretation.
The 25 native states stay tied to liquid fraction, its derived gas complement, instantaneous mixture velocity, thresholds, units, and coordinate conventions.
This case supplied requirements and representative external outputs for the adjacent Linux/headless workflow. VisualOpenFOAM has not yet run its real-case reader on these data.
VisualOpenFOAM documents the separate case-inventory, VTK, field-report, and headless Blender planning path. Its reader implementation remains separate from the scientific results here.
Interpretation and limits
The current laminar grid and 0.012 s window support phase-aware fields, connected-interface morphology, and reproducible visualization; they do not establish validated atomization, resolved breakup or spray, DNS/LES, production CFD, experimental agreement, or stationarity. Higher-resolution physics would require a separate HPC or adaptive-mesh campaign with planned sensitivity and validation work.