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

Optimization of a High Expansion ORC Turbine Using a Genetic Algorithm


Go-down orc2019 Tracking Number 143

Presentation:
Session: Session 1A: Turbines-Design Aspects (1)
Room: Olympia
Session start: 10:30 Mon 09 Sep 2019

Stephan Smit   s.h.h.j.smit@tudelft.nl
Affifliation: TU Delft

Quirijn Eppinga   Quirijn.Eppinga@triogen.nl
Affifliation: Triogen

Gustavo Otero Rodriguez   g.j.oterorodriguez@tudelft.nl
Affifliation: TU Delft

Rene Pecnik   r.pecnik@tudelft.nl
Affifliation: TU Delft


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

Abstract:

The expander of an organic Rankine cycle (ORC) turbogenerator is a critical component for the overall system performance. If a designer is able to increase the turbine efficiency by 10%, the overall cycle efficiency increases by 4%. However, methodologies and dimensionless diagrams used for steam and gas turbine design are unreliable for ORC turbines because (1) the expansion process takes place in the dense gas region, where the ideal gas law equation is not valid, and (2) the working fluid features a small enthalpy drop due to its molecular complexity, bringing about a large pressure ratio per turbine stage that, combined with the relative low speed of sound of the organic fluid, results in a highly supersonic flow in the stator outlet. This work presents an aerodynamic optimization, by means of a genetic algorithm, of a cantilever rotor of a single stage high expansion (∼100) ORC turbine, operating with toluene (C 6 H 5 CH 3 ) as a working fluid. The rotor blade geometry is defined by 18 geometrical parameters and is generated using an in-house blade design tool. The physical integrity of the blade geometry is assured by constraints on the minimal thickness of the blade. The resulting rotor geometry is evaluated using a Reynolds-averaged Navier-Stokes solver in a quasi-three-dimensional (Q-3D) framework to account for the increase of area over the rotor stage. A mixing-plane model connects the stator and rotor domains, making the simulation steady state. To model the non-ideal gas properties the EOS by Robinson et al. (1985) is used. The objective of the optimization is to minimize the total enthalpy at the outlet of the turbine stage. The optimized geometry successfully reduces a large flow separation in the rotor blade suction side which was present in the original design. As a result, the entropy production is reduced and the total-to-static efficiency has increased by 4.3%. In the future the design will be manufactured providing the possibility of assessing the performance of the blade and the corresponding flow solution.