Place for a photo (no lines around photo) Measurement methods for validation of time-averaged CFD modelling of CFBs Liekkipäivä 30.01.2014, Tampere, Finland Sirpa Kallio, Juho Peltola VTT Technical Research Centre of Finland Jouni Elfvengren, Jari Kolehmainen, Pentti Saarenrinne Tampere University of Technology CFD Modelling of Fluidization Processes Gas-solid flow in fluidized beds is dominated by fluctuations of velocities, solids concentration and chemical components. The effects of these fluctuations have to be included in the model! Alternatives: 1. Detailed transient simulations that resolve the fluctuations 2. Filtered modelling (time or spatial filtering) Real CFB Simulation results Detailed 04/02/2014 Filtered 2 The Filtered Approach Spatial filtering (coarse grid) and temporal filtering As the mesh resolution of a transient simulation decreases, the sub-grid closure models approach the steady-state models. At VTT focus is on time-averaged CFD modelling of fluidization Simulation of a long time period avoided Less sensitive to mesh resolution coarser mesh allowed Fast simulation of industrial furnaces possible 04/02/2014 3 Time-Averaged Simulation Approach Time-averaged form of the continuity and momentum equations used in the transient simulations for phase q are: q q q q Uq = 0 Uq Uq = M q q ( 1) q ( qs q 1) g q K gs u g p q us Time-averaged drag u 04/02/2014 U qu / q uq " uq Uq p qs q ps q q q u"q u"q ”Reynolds stresses” instantaneous velocity phase-weighted average velocity velocity fluctuation 4 Time-Averaged Simulation Approach: Model Development Based on high resolution, Eulerian transient simulations: All the terms in the momentum equation are individually averaged Evaluation of the relative importance of each term in different conditions Simulation data is used to develop of the closure models for the steady-state simulations >120 simulations have been carried out 04/02/2014 5 Time-Averaged Simulation Approach: Model Development Experimental work is required to support and validate the simulations! Both the transient and the time-averaged simulations Simultaneous measurement of velocity and concentration • A requirement for validation of time-averaged simulations! Satisfactory measurement methods are available for small scale 2D rigs 04/02/2014 6 Lab scale experiments at Åbo Akademi University Pseudo 2D units : CFB 04/02/2014 BFB 7 0.4 m X 0.015 m X 3 m CFB and 0.9 m x 0.015 m x 2 m BFB Pseudo 2D: Image-based velocity measurements Co-operation between VTT, Åbo Akademi and Tampere University of Technology Simultaneous measurement of instantaneous: Particle velocity: Particle Image Velocimetry (PIV), Particle Tracking Velocimetry (PTV) Solids volume fraction: Absorbtion of light, correlations of image Greyscale value Particle size distribution: Greyscale gradient direction matching with a circular mask Gray scale volume fraction estimate. 0.2 0.18 5 0.16 10 0.14 15 0.12 0.1 20 0.08 = 25 0.06 30 04/02/2014 0.04 0.02 35 5 10 15 20 25 30 35 40 45 50 0 8 Lab scale 2D units vs. industrial scale 2D lab scale Good visual access allows accurate optical methods Flow behavior is affected by the front and back walls Small size: entrance and exit effects affect most of the riser Industrial fluidized beds Extremely limited visual access Local measurements with probes Measurement methods differ from the ones used in 2D units Larger cold models Compromise between full scale and 2D More realistic flow fields than in 2D Better visual access than industrial scale 04/02/2014 9 Experiments at Chalmers cold CFB rig 0.7 x 0.12 x 8.0 m The bed material is glass beads. Measurements: Velocity and solids concentration with optical probe Pressure profile Circulation rate Velocity measurement with PIV was tested 04/02/2014 10 The Optical probe Provided by Chalmers Two fiber optical sensors Measures the amount of light reflected from particles Time delay between the signals can be determined by cross-correlating the signals Solids velocity measurement Emitter Detector Ideally: A high frequency measurement of solids velocity and volume fraction Usable in industrial conditions with cooling Calibration procedure of Rundqvist et al. 2003 for solids volume fraction 04/02/2014 11 Optical probe measurements at Chalmers Vertical alignment of sensors: Measures the vertical velocity. Time delay detection by cross-correlation Sampling: Frequency 5000 Hz, sampling: 512 blocks with 80% overlap. Normalized and median-filtered. The measured velocity can be offset by several factors: Too low solids volume fraction Large non-vertical velocities Very low velocities are not detected. If the suspension is uniform, velocity cannot be measured Very noisy data! 04/02/2014 12 Evaluation of measurement quality Fraction of successful measurements at three fluidization velocities. 04/02/2014 Examples of the velocity distributions: Measured (gray) Corrected by interpolation (black). 13 Effects of uncertainties on measurements Solids volume fraction measured with the optical probe: Full data (solid lines). Points with valid velocity data (dashed) 04/02/2014 Reynolds averaged solids measured with the optical probe Original data (solid lines) Corrected velocity distributions 14 Measured velocities and fluxes Favre averaged solids vertical velocity. 04/02/2014 Local solids flux measured with the optical probe 15 Measured solids velocity correlations Solids velocity correlation (< >) measured with the optical probe. 04/02/2014 Solids velocity correlation obtained by weighing with volume fraction. = , , 16 Qualitative effect of solids concentration detection threshold on the results The analysis is based on simulated data Velocity and voidage are strongly correlated Significant effects on time-averaged properties 04/02/2014 17 PIV measurements at Chalmers Center line of the bed was horizontally illuminated with a laser probe Dense falling clusters on the front wall complicate the analysis Complex pre- and postprocessing is required to obtain reliable results 04/02/2014 18 PIV measurements at Chalmers Experiments with fluidization velocities 2.0, 2.5 and 3.0 m/s @ 3.8 m height Velocity distributions were again corrected by linear interpolation 04/02/2014 19 PIV measurements at Chalmers Experiments with fluidization velocities 2.0, 2.5 and 3.0 m/s @ 3.8 m height 3.0 m/s was found too dense for credible PIV results 04/02/2014 20 Conclusions Measurement methods are associated with limitations and inaccuracies a combination of several measurement methods is necessary Optical probes (with dual sensors): Simultaneous measurements of solids volume fraction and the vertical velocity Usable at solids volume fractions above 0.01 Very noisy data Fluctuation measurement result are qualitative at best PIV measurements: Much more reliable lateral and vertical velocities over a region-of-interest Usable at solids volume fractions below 0.005 04/02/2014 21 Acknowledgements The authors gratefully acknowledge the financial support of Tekes, VTT Technical Research Centre of Finland, Etelä-Savon Energia Oy, Fortum, Metso Power Oy and Numerola Oy, and the support from Saarijärven Kaukolämpö Oy. The invaluable contribution of Filip Johnsson, David Pallarés and Ulf Stenman at Chalmers University of Technology and of Alf Hermanson, Debanga Mondal and Henrik Saxém at Åbo Akademi University are gratefully acknowledged. 04/02/2014 22 TECHNOLOGY FOR BUSINESS