Significant effort has been spent over the years to improve the accuracy and reduce the Computational Fluid Dynamics (CFD) simulation time required to predict performance for centrifugal compressors. Most of the emphasis has been on modeling the impeller and diffuser components. This paper presents an evaluation of volute modeling targeted at reducing simulation time while increasing the accuracy of the results. Providing accurate predictions of performance and operating range is critical to the equipment users as it allows reduction in design margins for plant equipment dependent on compressor performance (i.e. drivers, intercoolers and other auxiliary equipment).
The volute is the component that collects the flow from the diffuser and guides it into the discharge nozzle. Due to the circumferential variation (tongue) in the geometry, this component has to be modeled in its entirety (360 degrees); which results in very large grid sizes. The impeller and diffuser, normally modeled as a sector or pie slice, result in significantly smaller meshes. The volute models require large numbers of computing nodes to be solved and tend to have convergence issues.
The investigation, with the objective of reducing the amount of time required to run these simulations and improve the convergence of the runs, evaluated several mesh configurations that focused on grid density (element count), element aspect ratio and use of inflation layers. The domain evaluated consisted of several stationary components and one rotating component. The model started at the inlet guide vane section followed by the impeller, vanned diffuser, volute, and discharge nozzle. Commercial software ANSYS CFX was used to develop the meshes using tetrahedral/prism elements and complete steady-state CFD analyses.
Detailed flow field characteristics (total and static pressure, velocity streamlines, etc.) and key performance parameters (loss coefficient, pressure ratio, etc.) were compared for the various configurations evaluated. In addition, experimental measurements were used to validate the CFD results. The configuration that resulted in the shortest cycle time with the best performance accuracy was selected as optimum. Accuracy is paramount for performance prediction and reduction in simulation time will allow more volute iterations to be investigated, which would help improve volute performance in centrifugal compressors.