The aquisition of three dimensional flow measurements through a muli stage high speed axial flow compressor

Abstract

The focus of the thesis is research based around Cranfield University's 3-stage high speed axial compressor test rig. This newly built rig supported by European Commission funding has tested a set of conventionally stacked 2D rotor and stator blades. The results will be used to evaluate the CFD codes; a advanced three dimensional blade set will then be designed by a collaboration of companies. The aim of which is to prove a 2% efficiency gain for the advanced blade set. Due to the limited axial spacing between the measurement planes a unique seven probe assembly and traverse mechanism has been designed. The assembly has the ability to yaw the probes and control the insertion depths. This takes a number of different probes such as cobra probes, fast response (pneumatic) probes and temperature probes. A computer program has been written that completely automates the control of the probes and the data acquisition. This has allowed full area traversing of the compressor between rotor and stator blade rows. Due to the unique design of the seven probes, a high-speed wind turmel section has also been designed and manufactured. This has enabled the probes to be calibrated for mach numbers up to 0.78 as well as for both pitch and yaw. Due to the representative size, blade count and high flow forces of the high-speed compressor, instrumentation to capture the three-dimensional flow field in a high-speed environment has been researched. A small four-hole probe has been designed and used to take three dimensional steady state pneumatic measurements. This combined with a post processing program has provided very detailed results downstream of the stators through the compressor. This includes the complex three-dimensional flow structure and secondary flows associated with tip leakage, end wall boundary layer, wake transportation and blade row interactions. A fast response probe was designed and evaluated to capture the flow field downstream of the rotors but found to be insensitive to yaw angle. The results taken in a high-speed environment at full scale and engine representative speed supports some of the findings taken in a low speed environment. The author considers that the higher turbulence and speeds leads to increased mixing in the blade rows. This leads to almost all the ow being three dimensional in nature, this is not apparent in a low speed representation. The upstream wakes and their interaction with the downstream rows can be clearly seen. This is further demonstrated by the insertion of an upstream probe. The effects of this probe being transported axially downstream through a blade row. The results show the effectiveness and importance of a rigorous mapping procedure. This is particularly useful where the wake thicknesses are small and pressure gradients are high in comparison to the probe size. The size and extent of the areas of loss become more pronounced as results are taken through the compressor. This is true both at peak efficiency as well as near surge. The higher loading at the near surge condition increases these areas of loss still further.