Abstract: High Speed Compressor Design

A long-standing paradox exists in high speed compressor blade design. As the inlet flow to a row of blades approaches the speed of sound, the two-dimensional (2D) design of that blade row becomes exceptionally difficult, with designers laboriously trying to get to the one perfect blade shape that will both operate efficiently and maximise the useful incidence range between choke and stall. At the speed of sound, 2D design becomes virtually impossible. In practice however, despite design difficulties, it is possible to operate three-dimensional (3D) blade rows efficiently with regions close to the speed of sound (transonic) and above the speed of sound (supersonic) over a much wider incidence range than expected from 2D; even when the sectional blade designs are imperfect up the span due to for instance manufacturing error. This thesis finally explains this paradox with major implications in design.

Systems and processes for designing blades with an inlet speed much lower than the speed of sound (subsonic) exist. However, when the inlet velocity of a blade approaches the speed of sound (transonic) or exceeds it (supersonic) the engineering of high speed sections becomes more of an art rather than one based on a rigorous scientific design process; leading to bespoke blade designs. For this purpose, in this thesis, an automated blade design methodology is developed in the transonic and supersonic regime that robustly and consistently leads to the perfect aerodynamic shape for a given inlet Mach number and aerodynamic duty.

This for the first time allows the independent study of an aerodynamically perfect set of blades with increasing inlet Mach number. The key design parameter is physically derived as the area ratio: Athroat/Ainlet between the minimum flow area in the blade row passage, commonly known as the ‘throat’, and the inlet flow area upstream of the blade. Most surprisingly, approaching an inlet Mach number of unity only one unique value of the area ratio Athroat/Ainlet is shown to exist. If the area ratio deviates from this value, because of manufacturing error or a change in incidence, then the blade is choked or has its boundary layer separated due to a strong shock, making the design of the sonic streamtube virtually impossible.

The paradox is resolved due to a newly identified 3D mechanism termed ‘transonic relief’. A novel simple model is developed which allows ‘transonic relief’ to be decoupled from other mechanisms and be independently studied. Using this simple model, it is shown that the ‘transonic relief’ mechanism automatically readjusts the ratio of flow areas to its optimal value by redistributing the flow. Even though a redistribution of the flow is known to occur in transonic blades, this is the first time it is shown to always act to relieve the flow and the underlying physical mechanisms behind this are explained. As a result, the effects of ‘transonic relief’ are further linked to the key fundamental design parameters controlling the useful operating range of engine representative transonic blade rows, such as the hub to-tip ratio, and the results are translated to the preliminary design of high speed multistage core compressors.

Demetrios Lefas