combustion-cfd

name	combustion-cfd
description	Combustion CFD — turbulent combustion models (EDC, flamelet, PDF), reaction mechanisms (GRI-Mech), NOx/soot prediction, premixed vs. non-premixed, RANS/LES combustion, ANSYS Fluent setup.
metadata	{"priority":7,"promptSignals":{"phrases":["combustion CFD","combustion simulation","flame model","turbulent combustion","NOx prediction","flamelet model","EDC model"],"minScore":3}}

Combustion CFD — Complete Skill

Combustion Regimes

Premixed: fuel and oxidizer mixed before combustion (SI engine, gas turbine lean premix) Non-premixed (diffusion): fuel and oxidizer mix at flame (diesel, candle, gas turbine legacy) Partially premixed: intermediate; most practical combustors

Damköhler number: Da = τ_flow / τ_chemistry Da >> 1: fast chemistry (flame sheet); Da ~ 1: distributed reaction; Da << 1: no flame

Karlovitz number: Ka = (l_K/l_F)^2 (Kolmogorov scale / flame thickness) Ka < 1: thin flame (laminar flamelet valid) Ka > 100: broken reaction zone; chemistry-turbulence coupling important

Turbulent Combustion Models

Eddy Dissipation Model (EDM) / Eddy Dissipation Concept (EDC)

EDM (Magnussen & Hjertager): reaction rate limited by turbulent mixing R_f = A ρ (ε/k) min(Y_f, Y_ox/s) R_prod = AB ρ (ε/k) Y_prod / (1+s)

A, B = model constants (A=4.0, B=0.5 standard); s = stoichiometric ratio Simple; robust; poor for stiff chemistry; ignores kinetics

EDC (Extended): reaction occurs in fine structures (turbulent fine scales) γ*^3 = C_γ² ν ε^0.5 / k² (fine structure volume fraction; C_γ = 2.1377) Reaction allowed only in fine structures with local residence time τ*

EDC handles finite-rate chemistry; compatible with detailed mechanisms (10–50 species)

Non-Premixed: Mixture Fraction / Flamelet

Mixture fraction Z: Z=1 pure fuel; Z=0 pure oxidizer; Z_st = stoichiometric For Burke-Schumann: species mass fractions are functions of Z only (infinitely fast chemistry)

Flamelet model (Peters): Flame structure parameterized by Z and scalar dissipation χ = 2D|∇Z|² Flame library: precompute 1D flamelets across Z and χ → tabulate species + T χ_quench: flame extinguishes above this value; tabulate to extinction

PDF transport: solve for P(Z, Z'²) from transport equations Assume shape (β-PDF) or solve full PDF → weighted integration over flamelet table

Premixed: G-equation / Thickened Flame

G-equation (level-set): G=0 marks flame front ∂G/∂t + u·∇G = s_L |∇G|

s_L = laminar flame speed (from chemistry + turbulence wrinkling: s_T = Ξ s_L) Wrinkling factor Ξ from turbulence-chemistry interaction model

Thickened Flame Model (TFM): Artificially thicken flame to resolve on mesh; adjust reaction rates inversely E = thickening factor (10–100×); F = efficiency function for wrinkling Used in LES combustion (TFLES, AVBP code)

FGM (Flamelet Generated Manifold)

Tabulate all species from laminar flamelet solutions vs. progress variable c c = (T - T_u) / (T_ad - T_u) or c = Y_CO + Y_CO₂ Transport: c equation + Z equation + look up T, species from table Efficient; widely used in practical combustion LES

Reaction Mechanisms

GRI-Mech 3.0 (methane): 53 species, 325 reactions; validated for CH₄/air USCM (USC-Mech II): H₂/CO/C₁-C₄; 111 species LLNL (n-heptane): 544 species; diesel surrogate; very expensive in 3D CFD Reduced/skeletal mechanisms: from DRGEP (Directed Relation Graph with Error Propagation) Typical: 20–40 species sufficient for NOx/CO prediction

NOx Modeling

Thermal NO (Zeldovich mechanism): N₂ + O ⇌ NO + N (rate limiting) R_NO = 2k₁[O][N₂] / (1 + k₁[N]/(k₂[O₂]+k₃[OH])) k₁ = 1.8×10¹¹ exp(-38370/T) [m³/mol·s]

Temperature sensitivity: R_NO ∝ exp(-38370/T) → strong function of peak T Every 10K reduction → 15–20% NOx reduction at high T

Prompt NO (Fenimore): CH radicals react with N₂; significant for rich premixed N₂O pathway: important at low temperature, lean conditions

Post-processing approach: solve NO transport equation with frozen flow field Source term from Zeldovich + prompt correlation More efficient than solving NO in-loop

Soot Modeling

Two-equation (Moss-Brookes): solve soot volume fraction + particle number density Method of moments: track moments of soot particle size distribution Sectional: full PSD (particle size distribution) → most accurate; expensive

Primary factors: fuel-rich zone equivalence ratio φ >> 1 → soot precursors (C₂H₂, PAH) Temperature: peak soot at T ≈ 1500–1800 K; oxidizes above 1800 K

Fluent Combustion Setup

Model → Species → Species Transport (EDC or Premixed Combustion) Mechanism import: CHEMKIN format → reads .inp + .dat thermodynamic files Boundary: inlet temperature, equivalence ratio or species mass fractions Turbulence: k-ε or k-ω SST for RANS; LES-Smagorinsky for transient Radiation: P1 or Discrete Ordinates (DO) for sooty flames; gray gas absorption

Convergence: energy (10⁻⁶); species (10⁻⁵); velocity (10⁻⁴) Grid: fine near flame (< 0.1 mm for premixed RANS; < 0.5 mm for LES)

LES for Combustion

LES resolves large eddies; models sub-grid combustion with TFLES, FGM, or stochastic fields Better for:

• Blowout/ignition simulation
• Thermoacoustic instability
• Swirl combustor prediction

Cost: 100–1000× more than RANS; requires ≥ 10⁷ cells; parallel HPC

Thermoacoustic Instability

Pressure oscillations coupled to heat release → Rayleigh criterion: Q̃ p̃ dt > 0 (positive feedback → instability) Damped if combustor damping > Rayleigh driving

LES with acoustics → predict TA limit cycles; CFD/CEM hybrid methods

Output

Provide: combustion model selection (EDM/EDC/flamelet) with justification, Da and Ka assessment, mechanism size (species count), NOx prediction [ppm by volume] and primary contributor (thermal/prompt), peak temperature [K], flame length or stabilization point, convergence metrics, mesh requirements.