Fluid-dynamic characterisation of small-scale organic Rankine cycle radial-outflow turbine for renewable energy application

Abstract

Since the problem of global warming has been broadly considered as vital and due to constantly rising oil prices, further fossil fuels exploitation requires diametrical changes immediately. The demand for environmentally friendly improvements does not only entail searching for new resources but also utilizing those which are available and more importantly, renewable. Organic Rankine Cycle (ORC) is possibly the most flexible technology in terms of temperature level and capacity nowadays. It is often the only applicable means of conversion for external energy sources and is therefore a very active research area in the field. Project investigates performance of an unconventional, small-scale, three-stage radial outflow, ORC turbine as this component has proven to be critical in the system. Main content embraces steady and transient computational fluid dynamic simulations by means of Ansys CFX software and evaluation of loss mechanisms. The main project objective is to evaluate the performance of the machine at its design conditions and fluid-dynamic characterisation of the flow. Geometrical data has been obtained from TU Delft in-house mean-line code zTurbo for preliminary fluid-dynamic design of turbomachinery. The main findings of this work involve performance evaluation of an innovative mini-turbine configuration and the applicability of available loss estimation models. Although there are guidelines on designing conventional, larger, axial power plants, still little is known about research in the field of small-scale, radial outflow machines working with molecularly complex fluids. Overall total-to-static efficiency of the machine is 65.3% for a 0.1 mm tip leakage gap and sealed stators. It is higher, and in close proximity to the one of 63.1% obtained by mean-line prediction tool zTurbo. Thermodynamic processes undergone by the working fluid in mini-ORC ROT can be represented by velocity triangles at midspan with a 2.2% discrepancy in overall efficiency w.r.t. two-equation, steady RANS. Relatively uniform blade load, ability to apply tight clearance, tangential deflection in rotors due to Coriolis force and the relative motion w.r.t. the casing contribute to a decrease in 3D effects. This efficiency is expected to be slightly lower if tip leakage is imposed also on the stationary domains, possibly with even better match. Total pressure loss coefficient for the first stator, for the steady-state case without tip leakage is of exactly the same value (23%) as the one for transient simulation with free-slip endwalls, accounting only for two-dimensional effects, averaged over the oscillation period. For the rotor, where the endwall boundary layer is more disturbed, steady state case predicts this loss by 3 points higher w.r.t. transient case, which gives 28%. Together with small secondary losses predicted in the stator, accounting for only 14% of the overall loss, 2D estimation/optimisation methods can be possibly used to predict the performance. Naturally growing passage area, allowing for smaller flaring angle, contributes to a decrease in span-wise velocity components and diminishes 3D effects. Stage 1 proved to be more susceptible to tip leakage increase. Small aspect ratios, in the order of 0.5, in mini-ORC ramp up the impact of tip leakage vortex compared to larger blades. The poorest performance (total-to-total efficiency of 51.2%) has been noted for the third stage, which requires profile optimisation and stagger angle modification. Expectedly, increase in tip leakage gap in rotational domains, leads to rise in entropy, which is even slightly more perceivable in low-aspect ratio blades, such as Rotor 1. Good match has been found between CFD and zTurbo in terms of overall loss estimation. Discrepancies for Rotor 1, Stator 3 and Rotor 3 are 0.1%, 0.1% and 0.2%, respectively. For Stator 1, Stator 2 and Rotor 2 these discrepancies are higher, of 3.3%, 7.7% and 3.0%, respectively. The main design problems are: profile losses and 2D design, including shock interaction shown in the results from the transient simulation. It is expected that this work will make a long-standing contribution to the body of knowledge on loss mechanisms in small-scale ORC machines and it will help to build a more sustainable and cleaner world.