K-ε turbulence model

K-ε turbulence model
Name	K-ε turbulence model
Type	Eddy-viscosity model
Developer	Launder, Spalding
First appeared	1974
Equations	Two-equation model: turbulent kinetic energy (k) and dissipation rate (ε)
Applications	Engineering CFD, atmospheric boundary layer, combustion

K-ε turbulence model

The K-ε turbulence model is a widely used two-equation eddy-viscosity closure for Reynolds-averaged Navier–Stokes problems developed to represent turbulent transport via modeled equations for turbulent kinetic energy and its dissipation. It was introduced in the context of industrial United Kingdom research and has been adopted broadly across communities including NASA, U.S. Navy, European Space Agency, Siemens, and major academic groups at Imperial College London, Massachusetts Institute of Technology, Stanford University, and Princeton University. The model forms a cornerstone of many commercial and open-source codes maintained by organizations such as ANSYS, OpenFOAM, COMSOL, and CFD Research Corporation.

Introduction

The K-ε formulation originates from work by scholars at Birmingham and the original presentation by B.E. Launder and D.B. Spalding, who drew on earlier ideas from researchers at Prandtl-influenced schools and the statistical theory advanced in Reynolds analyses. It constructs transport equations for turbulent kinetic energy k and its scalar dissipation rate ε to close the Reynolds-averaged system, linking to an eddy-viscosity through an algebraic relation with k and ε. Adoption spread through industrial projects connected to Rolls-Royce, General Electric, and research programs at NASA Glenn Research Center and CERN for flow prediction tasks.

Mathematical Formulation

The standard K-ε model employs two partial differential equations: one for k (turbulent kinetic energy) and one for ε (dissipation rate), each containing production, diffusion, and sink terms. Constants introduced by Launder and Spalding are commonly cited alongside empirical coefficients derived through calibration against experiments from groups at NACA, Von Kármán-inspired wind tunnel campaigns, and datasets collected by National Physical Laboratory researchers. The modeled turbulent (eddy) viscosity μ_t is expressed as μ_t = C_μ ρ k^2/ε, connecting to mean-flow quantities in Navier–Stokes closures used in codes by NASA Ames Research Center and industrial partners like Boeing and Airbus. Wall treatments and low-Re modifications reference boundary-layer theory from studies at Princeton University and experiments at DTU Wind Energy.

Variants and Extensions

Multiple variants arose to address shortcomings: the realizable K-ε and RNG K-ε developed with contributions from NASA Langley and the Argonne National Laboratory, respectively, introduce modified transport terms and coefficients to improve behavior in strained and rotating flows used in projects by Lockheed Martin and Northrop Grumman. Low-Reynolds-number variants incorporate damping functions informed by work at Cambridge University and ETH Zurich. Hybridizations with Reynolds-stress models and large-eddy simulation approaches have been pursued by groups at University of California, Berkeley and Los Alamos National Laboratory, while combustion-specific extensions link to detailed mechanisms developed at Sandia National Laboratories and Imperial College London combustion labs.

Implementation and Numerical Aspects

Implementations appear in commercial and research solvers from ANSYS Fluent, STAR-CCM+, OpenFOAM Foundation, and bespoke codes at Lawrence Berkeley National Laboratory. Numerical stability depends on proper discretization (finite-volume, finite-element) and iterative solvers such as those from PETSc and Trilinos. Wall treatments use enhanced near-wall mesh strategies recommended in tutorials from Siemens PLM and boundary-layer experiments at Delft University of Technology. Time-stepping, linearization, and coupling with scalar transport or multiphase models have been refined through collaborations between MIT and ETH Zurich.

Applications and Performance

The model is applied across aerospace design at Boeing and Airbus, automotive aerodynamics at Ford Motor Company and Volkswagen, environmental flow modeling for agencies like EPA and Met Office, and process engineering in firms such as Dow Chemical and BASF. It performs well for fully developed turbulent shear flows including pipe flow, channel flow, and boundary layers validated against canonical datasets from Johns Hopkins University turbulence databases and experiments at Los Alamos National Laboratory. Modifications enable use in combustion chambers in research by Princeton Plasma Physics Laboratory and atmospheric boundary layer modeling used by NOAA and European Centre for Medium-Range Weather Forecasts.

Limitations and Criticisms

Critiques from turbulence researchers at Stanford University, Caltech, and ETH Zurich highlight deficiencies in separation prediction, mean-flow curvature, and strong anisotropy, especially in rotating or swirling flows encountered in GE Aerospace and Siemens Energy turbomachinery studies. The model’s isotropic eddy-viscosity assumption limits accuracy for flows dominated by Reynolds-stress anisotropy, prompting alternatives like Reynolds-stress models advanced by teams at University of Michigan and Imperial College London. Benchmarking campaigns organized by ERCOFTAC and inter-comparisons at CFD International conferences document systematic biases.

Practical Calibration and Constants

Standard constants (C_μ, C_1ε, C_2ε, σ_k, σ_ε) derive from calibration against experimental datasets collected by NACA, National Bureau of Standards, and university wind tunnels such as Delft University of Technology and University of Michigan. Practical calibration often occurs within industrial contexts at Rolls-Royce and Siemens where sector-specific adjustments and wall-function choices are informed by testing programs and validation suites maintained at NASA Glenn Research Center and Sandia National Laboratories.

Category:Turbulence models