Development of a 1-D performance prediction technique for automotive centrifugal compressors

The increasing demand for improved performance in diesel and petrol engines - particularly in the motor-sport industry - has increased the need for performance enhancing devices such as the automotive turbocharger. The prediction of compressor performance in the early design stage of a turbocharger is critical and helps to ensure that the range and matching of the constituent components (impeller, diffuser and volute) is satisfactory. Although the fluid flow inside the compressor is three-dimensional, effective analysis can be carried out using one-dimensional prediction techniques. Many prediction techniques have been developed over the decades and improvements to these methods have primarily been due to a greater understanding of compressor operation. The major gain from establishing more accurate prediction techniques is the reduction of uncertainties in both design time and production costs as well as allowing existing designs of centrifugal compressors to be improved. This thesis presents the work carried out to develop a PC-based I -D prediction technique called CAPRICE, with focus aimed at developing new models for the vaneless diffuser and volute casing. Extensive analysis of the existing models has been undertaken relating geometric features to performance. A specially-constructed interstage test rig, designed to extract data from the components, enabled experimental data from two compressors to be gathered. The data collected was used to develop several new correlations; A new correlation for the prediction of impeller work was produced and shown to be an improvement on the Wiesner equation. Formulae for the prediction of impeller and diffuser surge were developed which enabled a more accurate prediction of surge to be made. and diffuser and volute loss and recovery coefficient correlations were produced, separating the diffusion system model in CAPRICE for the first time. An equation for the prediction of the length of the log spiral path in the diffuser was also derived. The work was validated against existing I-D and 3-D models and shown to produce excellent comparisons and overall compressor maps have been produced to demonstrate these developments throughout this work. The resulting I -D performance prediction technique gives the designer better control of the overall perfon-nance by allowing a greater level of adjustment to be made to the individual component geometries, previously unavailable in CAPRICE.