Phase Doppler Anemometry in Unsteady Two-Phase Flow: A Novel Post-Processing Data Algorithm

Doktorsavhandling, 2006

This thesis presents a novel method to estimate particle fluctuations in unsteady two-phase flow using phase Doppler anemometry (PDA). A comprehensive derivation and a successful validation of a PDA data post-processing algorithm are given. Fluctuations in unsteady two-phase flow, such as in circulating fluidized-bed (CFB) boilers, are important to understand for prediction of combustion and generation and destruction of emissions. Thus, PDA measurements were performed in the upper dilute part of a riser of a cold laboratory-scale CFB, operated to approximately represent CFB boiler conditions. The existing method of processing the sampled raw time series (based on ensemble averaging) resulted in strongly biased moments of particle velocity and mass flux. On the contrary, in the new algorithm extensive properties of each validated particle are weighted with their inverse measurement volume to account for the procedure of particle sampling and fluctuations in the particle concentration, yielding unbiased time-moments. To compensate for the short characteristic length of the validation volume, the properties of particles are expressed by fields of particle groups, using a local averaging time. This allows for estimation of time fluctuations, which can be compared with results from numerical two-phase flow simulations, adopting an Eulerian/Eulerian approach. The novel algorithm was extensively tested and validated by applying it on: (i) artificial time series, which were generated based on a measured time series, (ii) a time series obtained by a numerical simulation of a gas/solid flow around a bluff-body, and (iii) measured time series covering a cross-section of the CFB riser, and comparison of the estimated vertical mass flux over the cross-section with its value in the cyclone leg. Particle clustering was investigated based on the PDA measurements in the CFB riser. A new cluster criterion was introduced to distinguish particle clusters from the surrounding dilute flow. Profiles of time mean and time fluctuations in particle velocity, concentration, and mass flux were estimated in the riser.