ANSYS posts cross‑injection Fluent demo

ANSYS just dropped a hands-on Fluent demo tackling one of CFD's trickier setups: jets injecting from opposing walls into a crossflow. This "cross-injection" mimics fuel sprays in gas turbine combustors or film cooling in turbine blades — where uniform mixing and penetration depth make or break efficiency. Engineers have wrestled with these for years because poor setup leads to lousy meshes, stalled convergence, or fake-looking flow physics. Today's 20-minute video from ANSYS's official channel walks you through a bulletproof workflow, from geometry to glyphs. ### What's cross-injection, exactly? Picture a duct with air blasting through — that's your crossflow. Now squirt jets perpendicular from the top wall and bottom wall, head-on into each other. The jets bend, shred, and mix under the crossflow's shear. This setup shows up in lean-burn combustors, where fuel must spread evenly without hot spots. Opposing injections boost uniformity but amplify turbulence — Fluent has to resolve counter-rotating vortex pairs without choking on mesh skewness. The demo uses a 2D channel slice, 1m long, with jets at 0.02m diameter. ### Why Fluent for this over other solvers? Fluent's strength here is its polyhedral meshing and robust wall treatments for high-velocity gradients. You get species transport for mixing fractions, plus DES or LES for unsteady vortical flow — critical since RANS often smears jet trajectories. The video picks Fluent's pressure-based solver with second-order schemes. No custom code needed; built-in jet-in-crossflow models handle the momentum flux right. Turns out, this scales to 3D annular combustors with minimal tweaks. ### How do you set up the injections? Start with a simple blockMesh-like domain in SpaceClaim — inlet velocity 50 m/s, jets at 100 m/s, air properties. Patch injection zones on walls as mass-flow inlets, not velocity inlets — that nails the choked flow physics. Set turbulence intensity to 5%, hydraulic diameter matching jet size. The key: enable "opposite wall" symmetry in boundary conditions to cut compute time without losing physics. Video shows the dialog boxes step-by-step; copy-paste values work for starters. ### What's the mesh strategy? Inflation layers rule here — 15 layers on walls with 1.2 growth rate, total thickness 0.001m to capture boundary layers. Prism layers near jets prevent orthogonality errors. Global size 0.005m, but refine to 0.0005m in shear zones using Fluent Meshing's curvature sizing. Poly-hexcore fills the bulk for efficiency — 2.5 million cells total, converges in 1000 iterations on a laptop. Skewness under 0.9 everywhere; the demo quality-checks with contour plots. ### How does visualization reveal the physics? Post-processing steals the show. Streamlines trace jet trajectories — top jet curls clockwise, bottom counterclockwise, merging mid-channel. Q-criterion isosurfaces (at 10^5 s^-2) light up vortex cores like neon signs. Liutex vorticity vectors highlight rotation without dilation noise — injectors penetrate 4-5 diameters before breakup. Mixing uniformity via mass fraction contours at 0.5 equivalence ratio; total pressure loss hovers at 3-4%. Animate it, and you see the horseshoe vortex dance. ### Why build this as a mini-project? Grab the video's.cas file (linked in description), tweak jet velocity to 150 m/s, and quantify penetration depth vs. momentum flux ratio — classic jet-in-crossflow parameter. Add a passive scalar for uniformity index, plot total pressure recovery along centerline. For propulsion folks, swap air for hot vitiated gas (premixed CH4-air), scale to real liner dimensions. Validates against experiments like GE's jet mixer data. Students: export to EnSight for that journal-ready PDF. ### Any gotchas to watch? Convergence stalls if under-relaxed turbulence; bump k-epsilon to 0.8. Jets can "bounce" off walls without sharp enough refinement — add local sizing on injection patches. For LES, time step at 1e-6 s, but start RANS to bootstrap. Video skips multiphase but nods to VOF for liquid sprays. Scales to Ansys 2025 R1 or later. Bottom line: This demo slashes the learning curve for combustor CFD from weeks to hours. Propulsion teams get production-ready mixing metrics overnight; academics, a validated baseline for papers. Fork it, iterate — next stop, your thesis sim. ```