PRELIMINARY DESIGN OPTIMIZATION OF AN ORGANIC RANKINE CYCLE RADIAL TURBINE ROTOR

The present study describes the application of a preliminary design approach for the optimization of an organic Rankine cycle radial turbine. Losses in the nozzle the rotor have initially been modelled using a mean-line design approach. The work focuses on a typical small-scale application of 50 kW, and two working fluids, R245fa (1,1,1,3,3,pentafluoropropane) and R236fa (1,1,1,3,3,3-hexafluoropropane) are considered for validation purposes. Real gas formulations have been used based on the NIST REFPROP database. The validation is based on a design from the literature, and the results demonstrate close agreement the reference geometry and thermodynamic parameters. The total-to-total efficiencies of the reference turbine designs were 72% and 79%. Following the validation exercise, an optimization process was performed using a controlled random search algorithm with the turbine efficiency set as the figure of merit: The optimization focuses on the R245fa working fluid since it is more suitable for the operating conditions of the proposed cycle, enables an overpressure in the condenser and allows higher system efficiency levels. The R236fa working fluid was also used for comparison with the literature, and the reason is the positive slope of the saturation curve, somehow is possible to work with lower temperatures. Key preliminary design variables such as flow coefficient, loading coefficient, and length parameter have been considered. While several optimized preliminary designs are available in the literature with efficiency levels of up to 90%, the preliminary design choices made will only hold true for machines operating with ideal gases, i.e. typical exhaust gases from an air breathing combustion engine. For machines operating with real gases, such as organic working fluids, the design choices need to be rethought and a preliminary design optimization process needs to be introduced. The efficiency achieved in the final radial turbine design operating with R245fa following the optimization process was 82.4%. A three-dimensional analysis of the flow through the blade section using computational fluid dynamics was carried out on the final optimized design to confirm the preliminary design and further analyze its characteristics.