Proceedings of the
5th International Seminar on
ORC Power Systems
9 - 11 September 2019, Athens Greece
Home Program Author Index Search

Robust Optimization of an Organic Rankine Cycle for Geothermal Application


Go-down orc2019 Tracking Number 164

Presentation:
Session: Session 5B: System design (2)
Room: Attica
Session start: 16:00 Tue 10 Sep 2019

Aldo Serafino   aldo.serafino@ensam.eu
Affifliation: ENERTIME / Laboratoire DynFluid, Arts et Métiers ParisTech

Benoit Obert   benoit.obert@enertime.com
Affifliation: ENERTIME

Léa Vergé   lea.verge@enertime.com
Affifliation: ENERTIME

Paola Cinnella   paola.cinnella@ensam.eu
Affifliation: Laboratoire DynFluid, Arts et Métiers ParisTech


Topics: - System Design and Optimization (Topics), - Simulation and Design Tools (Topics), - Oral Presentation (Preferred Presentation type)

Abstract:

A high level of know-how has been reached about Organic Rankine Cycles (ORCs). However, improving and optimizing their design is still a fundamental issue. Traditionally, thermodynamic and techno-economic ORC optimization have been performed only at the point identified by plant nominal working conditions. Just recently, some optimizations have started to consider also part-load performance. In any case, the approach used in all these works is always deterministic, as all model hypothesis and operating conditions are considered as a priori perfectly known. In practice, due to the manifold sources of uncertainty, the determistic approach is not thermodynamically and economically efficient and it is not always optimal for part-load operation. To overcome these difficulties, in this work we adopt instead a Robust Design (RD) for the design of ORCs under uncertainty. Specifically, an innovative RD optimization (RDO) technique relying on two nested Bayesian Kriging surrogates is employed to perform a thermodynamic cycle optimization by means of Taguchi’s RD criteria. The objective is to design an efficient ORC for geothermal application, affected both by epistemic uncertainty, mainly generated by the unknown properties of the thermal source, and aleatory one, given by the variable condensing temperature. The proposed RDO approach is used to select the best values for the design parameters avoiding an excessive sensitivity of the system to changes in the nominal operating conditions. The optimal solution obtained with the proposed approach is compared with the results from the deterministic optimization of the problem.