Optimization of experiment span and data acquisition rate for reliableelectrical capacitance tomography measurement in fluidization studies—a casestudy

Abstract

In a conventional fluidized bed, the particle behaviour is highly unpredictable andtherefore the data collected during a specific experimental scenario is often highlyrepresentative of that isolated result, rather than being reproducible. To quantifythe degree of representation/reproducibility, different scenarios of sampling spanand measurement frequency were tested for studying the bed hydrodynamics. Theexperiments were carried out at ambient conditions in a non-reacting fluidizedbed. A twin plane electrical capacitance tomography system was used to measurethe solid fraction distribution in a 150 mm diameter acrylic column and at aselected gas velocity of 0.9 m s−1. The images produced from capacitancemeasurements are relatively low-resolution images. Thus an iterative methodbased on a LPB algorithm has been used and the recommended number ofiterations required for enhanced images is presented. In measuring thehydrodynamic parameters, it is demonstrated that increasing the recording spanconsiderably increases the measurement accuracy. It is also observed that thevisualization of the axi-symmetrical nature of the bed hydrodynamicsis not easy to achieve with a short recording span of 20 s, due to thehigh bed non-uniformity. For solid fraction measurements, the minimumrecommended experimental sample is 4000 data points. When measuringdynamic parameters such as frequencies or standard deviation of solidfraction fluctuations, high data capture rates are of vital importance toproperly characterize the hydrodynamic fluctuations. Local measurementswere found to only represent the specific measuring location. In orderto have a global estimate of a dynamic parameter such as the bubblefrequency the analyses should be based on the average bed signal. Whenmeasuring the bubble rise velocity from the average bed signal it has beenfound that, within the operating conditions considered in this study, acombination of 60 Hz and 80 s (4800 points) or 100 Hz and 60 s (6000 points)produces an acceptable level of accuracy. Low data capture rates ≤40Hz may dramatically underestimate or completely fail to provide an estimate ofbubble rise velocity.