There is a growing interest in organic Rankine cycle (ORC) turbogenerators because of their ability to efficiently utilize external heat sources at low-to-medium temperature in the small-to-medium power range. High-temperature ORC turbines typically operate at very high pressure ratio and expand the organic working fluid in the dense-vapor thermodynamic region, thus requiring computational fluid dynamics (CFD) solvers coupled with accurate thermodynamic models for their performance assessment and design. In this paper we present a steady-state three-dimensional viscous CFD study of the Tri-O-Gen ORC radial turbine, including the radial nozzle, the rotor and the diffuser. The turbine operates with toluene as the working fluid, whose accurate thermophysical properties are obtained with a look-up table approach. Based on the 3D simulation results, together with a 2D fluid dynamic optimization procedure documented elsewhere, an improved nozzle geometry is designed, manufactured and experimentally tested. Measurements show it delivers 5 kW_e or 4% more net power output, as well as improved off-design performance.

@en

This paper presents the numerical investigation of real-gas effects occurring in the nozzle of an existing axial organic Rankine cycle turbine. Two-dimensional Reynolds-Averaged Navier- Stokes simulations are carried out using a computational fluid dynamics code which employs accurate models for the computation of the thermodynamic and transport properties of the working fluid. Two cases have been analyzed: The expansions starting from the subcritical conditions corresponding to the nozzle design point, and an expansion starting from supercritical conditions. The computations have been performed with two different thermodynamic models, the simple polytropic ideal gas equation of state (EoS) and a state-of-the-art multiparameter Span-Wagner EoS. For the computation of transport properties, two models have been employed, namely, a simple power law in the temperature and the more accurate (and complex) Chung model. The influence of the different fluid models on the computed flow field are discussed.@en