| description abstract | An important question for a designer is how, in the design process, to deal with the small geometric variations which result from either the manufacture process or inservice deterioration. For some blade designs geometric variations will have little or no effect on the performance of a row of blades, while in others their effects can be significant. This paper shows that blade designs which are most sensitive are those which are susceptible to a distinct switch in the fluid mechanisms responsible for limiting blade performance. To demonstrate this principle, the sensitivity of compressor 2D incidence range to manufacture variations is considered. Only one switch in mechanisms was observed, the onset of flow separation at the leading edge. This switch is only sensitive to geometric variations around the leading edge, 0–3% of the suction surface. The consequence for these manufacture variations was a 10% reduction in the blade's positive incidence range. For this switch, the boundary in the design space is best defined in terms of the blade pressure distribution. Blade designs where the acceleration exceeds a critical value just downstream of the leading edge are shown to be robust to geometric variation. Two historic designs, supercritical blades and blades with sharp leading edges, though superior in design intent, are shown to sit outside this robust region and thus, in practice, perform worse. The improved understanding of the robust, region of the design space is then used to design a blade capable of a robust, 5% increase in operating incidence range. | |
