A study of the internal flow of dense vapours used in Organic Rankine Cycle (ORC) turbines
File(s)
Author(s)
Chen, Tao
Type
Thesis or dissertation
Abstract
An Organic Rankine Cycle (ORC) is a thermodynamic cycle utilizing a heat source at low-temperature. It can be used for the waste heat recovery of vehicle engines, industrial processes, solar thermal power plants, and geothermal power plants, leading to reduction of CO2 emissions. The turbine expander is a key component of this cycle, and its efficiency is critical to the system performance. To improve the design of an ORC turbine, the internal flow of the turbine should be studied. Unlike the fluids in conventional turbines, the fluids used in ORC turbines are dense vapours. These vapours have complex molecules and relatively large molecular weights, and they operate at states close to thermodynamic critical points or saturation line. Therefore, the thermodynamic behaviours of dense vapours are far from those of ideal air. However, the influence of these effects on the internal flows is not well understood. The fundamental flow behaviours of dense vapours, including gasdynamic behaviours in blade-shaped nozzle flows and turbulent behaviours in wall-bounded flows, are the focus of this work.
A supersonic cascade using R1233zd(E) as working fluids is designed by the method of characteristics. The designed geometry is
able to achieve a nearly uniform outlet flow at about $Ma=2$, which is checked in an RANS simulation. A blade-shaped nozzle is designed using the blade shape of this cascade, and the preliminary test results of this nozzle is presented. This nozzle is tested with both nitrogen and R1233zd(E) as working fluids. In the test with nitrogen, an adverse pressure gradient is measured on both nozzle surfaces downstream of the nozzle throat, and a shock train is observed at the corresponding position. Similar to the nitrogen test, the adverse pressure gradient is also found in the tests with R1233zd(E), but the Schlieren images cannot clearly show the shock train due to the disturbance of two-phase flows.
A Direct Numerical Simulation (DNS) method for dense vapours is developed to obtain detailed information on turbulence. A modified Steger-Warming splitting is proposed to consider the strong non-ideal effects of gases, and the Span-Wagner EoS \cite{span2003equations} is used for studied dense vapours. The first studied benchmark case of wall-bounded flow is the supersonic fully-developed channel flow. Both the mean flow fields and the turbulent fluctuation fields are analysed. The mean profile and the fluctuation of thermodynamic properties are significantly affected by both molecular-complexity effects and non-ideal effects, and the sound-wave mode can be the dominant mode for generating fluctuations of thermodynamic properties ($T'$, $\rho'$) in dense vapours. Important modelling issues for dense vapours are also discussed, including the Strong Reynolds Analogy (SRA) and assumptions required for the $k-\varepsilon$ RANS method. The second studied benchmark case of wall-bounded flow is the bypass laminar-turbulent transition over a flat plate under a supersonic incoming flow. At the same incoming non-dimensional numbers ($Ma_\infty$ and $Re_x$), the breakdown of laminar flow starts earlier in dense vapours than in air. The mechanism of breakdown for both dense vapours and air is due to the interaction of two streamwise vortices in opposite rotational directions, leading to Kelvin-Helmholtz (KH) instability in a high shear layer. Proper Orthogonal Decomposition (POD) is also used to support the findings in vortex structure analysis.
A supersonic cascade using R1233zd(E) as working fluids is designed by the method of characteristics. The designed geometry is
able to achieve a nearly uniform outlet flow at about $Ma=2$, which is checked in an RANS simulation. A blade-shaped nozzle is designed using the blade shape of this cascade, and the preliminary test results of this nozzle is presented. This nozzle is tested with both nitrogen and R1233zd(E) as working fluids. In the test with nitrogen, an adverse pressure gradient is measured on both nozzle surfaces downstream of the nozzle throat, and a shock train is observed at the corresponding position. Similar to the nitrogen test, the adverse pressure gradient is also found in the tests with R1233zd(E), but the Schlieren images cannot clearly show the shock train due to the disturbance of two-phase flows.
A Direct Numerical Simulation (DNS) method for dense vapours is developed to obtain detailed information on turbulence. A modified Steger-Warming splitting is proposed to consider the strong non-ideal effects of gases, and the Span-Wagner EoS \cite{span2003equations} is used for studied dense vapours. The first studied benchmark case of wall-bounded flow is the supersonic fully-developed channel flow. Both the mean flow fields and the turbulent fluctuation fields are analysed. The mean profile and the fluctuation of thermodynamic properties are significantly affected by both molecular-complexity effects and non-ideal effects, and the sound-wave mode can be the dominant mode for generating fluctuations of thermodynamic properties ($T'$, $\rho'$) in dense vapours. Important modelling issues for dense vapours are also discussed, including the Strong Reynolds Analogy (SRA) and assumptions required for the $k-\varepsilon$ RANS method. The second studied benchmark case of wall-bounded flow is the bypass laminar-turbulent transition over a flat plate under a supersonic incoming flow. At the same incoming non-dimensional numbers ($Ma_\infty$ and $Re_x$), the breakdown of laminar flow starts earlier in dense vapours than in air. The mechanism of breakdown for both dense vapours and air is due to the interaction of two streamwise vortices in opposite rotational directions, leading to Kelvin-Helmholtz (KH) instability in a high shear layer. Proper Orthogonal Decomposition (POD) is also used to support the findings in vortex structure analysis.
Version
Open Access
Date Issued
2022-03
Date Awarded
2022-11
Copyright Statement
Creative Commons Attribution NonCommercial NoDerivatives Licence
Advisor
Martinez-Botas, Ricardo
Yang, Bijie
Publisher Department
Mechanical Engineering
Publisher Institution
Imperial College London
Qualification Level
Doctoral
Qualification Name
Doctor of Philosophy (PhD)