Multi-fidelity robust design optimization of an ORC turbine for high temperature waste heat recovery - Archive ouverte HAL Accéder directement au contenu
Article Dans Une Revue Energy Année : 2023

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

Aldo Serafino
Benoit Obert
  • Fonction : Auteur
Paola Cinnella

Résumé

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

Dates et versions

hal-03940140 , version 1 (15-01-2023)

Identifiants

Citer

Aldo Serafino, Benoit Obert, Paola Cinnella. Multi-fidelity robust design optimization of an ORC turbine for high temperature waste heat recovery. Energy, 2023, pp.126538. ⟨10.1016/j.energy.2022.126538⟩. ⟨hal-03940140⟩
5 Consultations
0 Téléchargements

Altmetric

Partager

Gmail Facebook X LinkedIn More