Study of macro-instabilities in stirred tanks using a velocity decomposition technique

In this paper a quantitative measure of errors introduced in the turbulent velocity RMS signal due to the presence of macroinstabilities (MI) in the velocity field is presented. The velocity time series were measured for four commonly used impellers (PBT, A310, HE3 and RT) with a one component LDV. Two locations in the tank, the impeller stream I and upper corner U, were studied. Three aspects of the geometry were varied: impeller diameter ( D = T/2 and D = T/4 ); number of baffles (two and four); and off-bottom clearance ( C/D = 1.0 and C/D = 0.5). By resampling and smoothing the velocity records the RMS velocity due to MI, υ MI , was determined. Further velocity decomposition recovered the high frequency component of the signal, υ HF , due to the randomfluctuations and the blade passages. Inclusion of the non-stationary, non-equilibrium MI component in the calculation of the RMS velocity can result in an overestimation of up to 50%. Analysis of the time series records shows that the MI is present in all configurations tested. In some cases (PBT and RT) the MI dominates the signal while for others (HE3 and A310) the amplitude of the signal is low and the MI is much less pronounced. The MI is very sensitive to geometry: for the impeller stream of the RT, increasing the number of baffles from two to four completely changes the velocity time series. The MI can have as dramatic an impact on experiments and analysis of the flow as the trailing vortices observed behind impeller blades. An understanding of this phenomenon is important for accurate analysis of the velocity and turbulence fields, for measurement of blend time and solids distribution, and for improving understanding of mesomixing.