Development of experimental and computational techniques for investigation of a heated pulsed air jet

Abstract

The temperature and velocity fields of a fully pulsed axisymmetric heated air jet have been investigated to determine its heat and mass transfer characteristics. Measurements were made using hot-wire anemometry, cold-wire resistance thermometers and laser Doppler anemometry (LDA). New techniques for hot- and cold-wire anemometry were developed and consideration was given to errors associated with LDA when used with high shift frequencies. Both the unheated and heated pulsed jets were simulated using the k-e turbulence model. The commercial partial differential equation solver PHOENICS was used to simulate the fully pulsed jet together with the k-e turbulence model. Many of the features of the pulsed jet which were observed experimentally were simulated successfully at least qualitatively. These include flow reversal near the edge of the jet, increased entrainment when compared with steady jets and large radial outflow near the leading edge of the pulse and radial inflow near the edge of the jet. The results showed that the pulsed component dominated up to about x/d = 50. Transition from pulsed jet to steady jet behaviour shown experimentally by a kink in the axial velocity decay curve could not be reproduced by the simulation. When (LDA) is used in a flow such as the pulsed jet where one velocity component is much larger than the other components, high shift frequencies are required in order to avoid fringe bias. This can lead to significant errors in the small velocity components if the timing of the burst envelope generated in the signal processor is incorrect. Tests with signal generators and in a steady jet flow showed that the LDA processors used for the present work had a timing error of approximately half a clock count. This error could be corrected either by software corrections to the resultant velocities or by a hardware adjustment to the processors. An investigation was also made into the response of inclined hot-wires to fluid temperature changes over a range of incident angles. The change of the wire voltage with fluid temperature was found to be a function of only the effective cooling velocity, allowing implementation of a simple temperature compensation technique suitable for both analogue and digital data processing. Open-loop compensation using a zero-pole network is often used to extend the frequency response of cold-wire resistance thermometers. For periodic flows such as the pulsed jet, the time constant of the wire changes significantly throughout the cycle and hence a novel velocity dependent system where the time constant of the compensator matches that of the wire was developed. A locally developed calibration interpolation technique for X-array hot-wire probes was extended and modified for X-probes with 120° between the wires. When used in an unheated fully pulsed jet the large reversed flow and contamination by the azimuthal velocity component led to the Reynolds shear stress and the radial component of the velocity fluctuation significantly lower than those obtained with the LDA. Pulsed jet temperature field results showed domination by the pulsed component up to x/d = 50 to 60 as reported previously for the axial velocity. Simultaneous measurements of axial velocity and temperature with a two-wire probe showed that the velocity-temperature covariance, uδ contained a significant proportion of the axial heat flux. From these data, calculations also showed that the vδ covariance was significantly higher than for steady jets. The turbulent Prandtl number was calculated to be in the range 0.6 - 0.7 for the inner part of the jet.