A new method for decomposition of high speed particle image velocimetry data

A high speed particle image velocimetry (high speed PIV) system developed by the National Energy Technology Laboratory (NETL) is being applied to measure individual particle motion in flow fields of high particle concentration. Particle flow fields were measured in two risers of circulating fluidized bed (CFB) systems, one riser of 0.203 m (8 in.) diameter with 80 μm mean diameter FCC particles flowing at velocities up to 30 m/s and one riser of 0.305 m (12 in.) diameter with 800 μm mean diameter HDPE particles at velocities up to 15 m/s. Particle concentrations ranged from zero to maximum packing in particle clusters. In these risers the high speed PIV system achieves sustained data rates of 0.1 to 1.0 million velocity vectors per second. This produces time series data for particle velocity that measures the full temporal range of velocity fluctuations. For comparison with CFD models that decompose particle velocity into a mean and a random fluctuating component of particle velocity in a manner similar to Reynolds Decomposition of single phase flows, the particle velocity data must be decomposed into a non-stationary mean component and a random component. The standard Reynolds decomposition method, which utilizes ensemble averaging, is inadequate for this application because particle velocity is under-sampled when particle concentration is low. We present a local window averaging method that decomposes the particle velocity time series even when particle velocity is being under-sampled due to periods of low particle concentration. This method decomposes particle velocity accurately and without loss of high frequency components of the velocity signal. Implementation of this method has led to the first measurements of the random component of particle velocity (and parameters derived from it, such as granular temperature) in a CFB riser that detects the entire temporal range of the particle velocity time series.