Abstract Hydrocarbon and nuclear fuels will contribute substantially to the growing global energy portfolio for some time to come. At the same time atmospheric carbon levels and economic concerns are increasingly shaping the power generation environment. Supercritical fluid technologies are being investigated to address both concerns. For example, direct fired supercritical CO 2 power cycles offer key advantages for carbon capture and thermal efficiency. Additionally, the supercritical water reactor Generation IV nuclear reactor allows higher power densities and more efficient operation than current pressurized or boiling water reactors. As these technologies are developed, a need exists for accurate, yet cost-effective modeling and simulation solutions to assist with design and safety calculations. Supercritical fluid properties are strongly dependent on temperature which results in a number of unique flow features. Among these features are the complex buoyancy and buoyancy-turbulence interaction behaviors exhibited by supercritical flows that are not found in liquid or gas flows. Therefore, standard turbulence models with current model coefficient values are not likely to be applicable to supercritical fluid flows. The subject of this paper is efficient modeling and simulation of supercritical flows to support nascent technologies growing to maturation and operational deployment. We present a framework for developing Reynolds-Averaged Navier-Stokes turbulence models specifically equipped for the challenges of supercritical flows. A novel formulation of the algebraic heat flux model of the buoyancy production of turbulence term is used with a traditional shear stress transport model. To produce a new turbulence model for supercritical channel flows, the empirical coefficients of the resulting four equation model were calculated from data previously published by other authors regarding upward supercritical flow through heated pipes. The presented SST k t - ω t approach was validated against heated tube experiments to show predictive capabilities in moderate flow conditions, where buoyancy effects are important but not dominating of inertial effects.