Published online by Cambridge University Press: 20 April 2006
The evolution of temperature variance and heat flux in decaying grid turbulence with a linear cross-stream temperature gradient is studied by producing the temperature gradient by means of two different methods: (a) by placing a ‘mandoline’ (Warhaft & Lumley 1978) downstream from the grid but with its wires differentially heated for the present study, and (b) by differentially heating ribbons of nichrome (a ‘toaster’) placed in the plenum chamber of the wind tunnel. For the former method the initial thermal/mechanical lengthscale ratio Lθ/L was varied by changing the mandoline configuration. For this method it is shown that the gradient causes Lθ/L to equilibrate to a value of about 0·9 regardless of its initial value, and that when this value is achieved the temperature variance increases approximately linearly with time. The toaster was used to produce a temperature gradient without the associated initial temperature variance (and initial thermal lengthscale) that is necessarily produced by the mandoline wires; for the toaster the temperature variance was produced solely by the action of turbulence against the temperature gradient. For this experiment too, the thermal variance grew linearly with time, and Lθ/L was approximately the same as the equilibrium value for the mandoline experiments. The equilibrium value of the ratio of temperature-variance production to temperature-variance dissipation was approximately 1·5 for all of the experiments. The ratio of the mechanical-dissipation/thermal-dissipation timescales was also found to equilibrate, but there was considerably more scatter in the data for this parameter. The values of the equilibrium length- and timescale ratios were not affected by the magnitude of the temperature gradient, which was varied for both experiments. Good transverse homogeneity in the thermal field was achieved in all cases, in contrast with previous experiments (using heated grids).