Multi-fidelity robust design optimization of an ORC turbine for high temperature waste heat recovery

The ORC technology is subject to manifold sources of uncertainty that can have a severe impact on the thermodynamic and economic efficiency plant components, particularly when system operated at off-design conditions. In this contribution we focus development turbines with stable performance under uncertainty: novel multi-fidelity robust design optimization (RDO) strategy used first nozzle an turbine for high temperature waste-heat recovery. For kind application, inlet outlet conditions may vary randomly over large range. RDO combines parsimonious quantification techniques multi-objective genetic algorithm optimizer based surrogate models. approach allows estimate accuracy low computational cost statistical moments probability distribution function quantity interest, which here entropy generation within cascade. To improve model coupled optimizer, expected improvement criterion adopted. converges efficient optimum solution, ensuring improved whole considered range uncertain operating significantly lower than other approaches proposed in literature.