An unstructured Tablulated Method for the computation of Thermo-physical fluid properties

Abstract

In various propulsion and power systems, modeling of non-ideal fluid flows (fluids that depart from ideal gas behaviour), presents a great challenge. For example, in organic rankine cycle (ORC) turbines, where a part of the expansion process occurs close to the vapour saturation curve, the flow deviates highly from ideal behavior. A branch of fluid dynamics called the Non-ideal compressible fluid dynamics (NICFD) deals with the modeling and analysis of such non ideal fluid flows.

As a consequence of the need for accurate thermo-physical property computation, various models have recently been developed for non ideal flows and a number of libraries are available to accurately predict the thermo-physical properties. However, the available thermodynamic libraries are often computationally costly since they require solving of complex equations of state (EoS) to obtain thermo-physical properties. When these libraries are coupled with existing simulation codes, (for example in computational fluid-dynamics), the simulation process is computationally inefficient.
This thesis is an endeavor towards enhancing the computational efficiency of the process of thermodynamic property calculation with the use of the Look up table (LUT) approach.

The LUT method aims at computing thermodynamic properties of a fluid with the help of array indexing operations applied on pre calculated or existing thermodynamic tables. These tables are initially obtained from a thermodynamic library FluidProp. A binary search algorithm helps in accurately locating the query point(s) on the thermodynamic domain. A data interpolation algorithm is then used to predict the thermodynamic properties of interest. The presented LUT method ensures inherently high accuracy with a very small computational cost, as demonstrated later in the thesis.

To check the applicability of the LUT method, it is used to obtain the pressure variation across a control volume with subsonic flow conditions. As a second and a much larger application, the LUT tool is coupled with an in house MOC (Method of Characteristics) tool to design the geometry of a supersonic nozzle.

A comprehensive analysis of this method is presented by comparing the accuracies and computational cost with the results from FluidProp. Both interpolation methods implemented in the proposed LUT method prove to be computationally efficient and accurate. The method is successfully applied to the MOC tool to design the geometry of the diverging section of a supersonic nozzle.