Elucidating Bubble Deformation in Fluidized Beds with Vertical Tube Banks Using X-Ray Tomography and Optical Probes

Riechmann, P., Paul Scherrer Institut
Schildhauer, T. J., Paul Scherrer Institut
Schillinger, F., Paul Scherrer Institut
Wagner, E. C., Delft University of Technology
Mudde, R. F., Delft University of Technology
van Ommen, J. R., Delft University of Technology

Proper understanding of the hydrodynamics in fluidized beds is the basis for successful design and safe scale-up of such reactors. This also holds for bubbling fluidized beds for strongly exothermic reactions such as methanation reactors in Power-to-Gas applications for energy storage. In such reactors, immersed heat exchanger tubes allow removing the heat of reaction which avoids the formation of hot spots and favors higher conversions in this equilibrium limited reaction system [Schildhauer2016].Voids or bubbles play a key role: they have a positive impact on the heat transfer and the movement of catalyst particles which was shown to avoid catalyst deactivation by managing the surface coverage with carbonaceous species [Seemann2010]. On the other hand, higher gas flow rates and the inherently large bubbles can lead to mass transfer limitations [Beetstra2009] and the increase of particle elutriation [Chew2015].

To better understand the influence of heat exchanger tubes on the size, growth rates and rise velocity of bubbles, Paul Scherrer Institut and TU Delft applied a number of methods in cold flow models and pilot scale, pressurized reactors with Geldart B type materials. This way, it could be shown that vertical heat exchanger tubes avoid the local defluidisation observed between dense banks of horizontal tubes [Schillinger 2015]. Further, vertical tubes structure the fluidized bed such that the local hydrodynamics is independent of the bed diameter [vanOmmen2007, Rüdisüli2012b]. Therefore, the sectorial scale-up approach enables to maintain the hydrodynamic situation when the reactor is scaled up by increasing the number of heat exchanger tubes (with constant tube diameter) with the square of the bed diameter ratio [Rüdisüli 2012a, Maurer 2014].

Pressure fluctuation measurements showed that bubbles grow significantly less in reactors with vertical tubes than in beds without immersed internals and also more slowly than expected from common correlations in literature [Karimipour2011]. This highlights the fact that the immersed heat exchanger tubes cause dominant “wall” effects for which reason none of the typical equations an correlations for fluidized beds can be used a priori for beds with immersed heat exchanger tubes.

Bubble hold-up and bubble distribution in the bed as well as bubble sizes and rise velocities were investigated in detail in beds of different diameters and vertical tubes arrangements [Schillinger 2017]. Both optical sensors and X-ray tomography were applied. The former method can be applied in hot and pressurized systems, but delivers information of the local situation at the sensor tip only while the latter gives full insight of the situation over the complete cross section of the bed [Maurer 2016]. It was shown that square and concentric arrangements of vertical tube banks deliver similar results and that different bed diameters lead to nearly identical bubble properties if the number of tubes was adapted according to the sectorial scale-up approach [Schillinger 2017].

The strictly local information obtained by optical sensors has to be interpreted correctly with respect to statistical effects, because optical probes have to be used at several radial positions and pierce large bubbles with a higher probability than small bubbles, but do not necessarily pierce them at the largest chord length. It has been shown by Monte-Carlo simulations that these two errors balance out for spherical bubbles yielding in close to correct average bubble diameters [Rüdisüli 2012c]. Already for the bubble rise velocity, the higher piercing probability for larger bubbles causes a bias in the measurements which necessitates the use of statistical methods to derive the original average bubble rise velocity [Maurer 2105].

For chemical reactors, average bubble sizes and rise velocities are however not sufficient to predict the reactor performance because e.g. the mass transfer will be limited most by large and fast bubbles. Therefore, knowledge of the distributions of bubble sizes and rise velocity are needed. Moreover, none of the above mentioned statistical methods work a priori, if the bubbles are not spherical as it is the case in beds with immersed internals.

The results from a fictitious optical sensor signal synthesized from the X-ray data were compared to the results of the 3D reconstruction (two spatial dimensions in the cross section, one time dimension) based on X-ray tomography data [Schillinger 2018]. Interestingly, when the bubble rise velocities are mapped as function of the measured geometrical dimension (pierced chord lengths for optical sensors, volume-equivalent diameters for the X-ray data evaluation), both methods seem to give deviating messages. Optical sensors measure high bubble rise velocities for bubbles with high pierced chord lengths which fits to typical bubble rise velocity correlations where the rise velocity increases with the square root of the bubble diameter . Evaluation of the reconstructed X-ray data show, however, smaller velocities for large-volume bubbles and both fast and slow bubbles with smaller volume, but hardly fast bubbles with large volume [Maurer 2016, Schillinger 2017].

The seeming inconsistency of both methods could now be explained by plotting the rise velocities of bubbles as function of their aspect ratio (largest cross section compared to maximum length in axial direction). While fast bubbles are slender with high aspect ratios (leading to high pieced chord length even with small volume), slow bubbles have a tendency to low aspect ratios which causes smaller pierced chord lengths even at high bubble volumes.

This shows that in systems far from standard assumptions such e.g. spherical bubbles, strong deviations from common correlations have to be expected which necessitates the combined use of several measurement methods to obtain the full picture. This then opens a pathway to derive design approaches also for non-standard reactor systems.

[Beetstra2009] Beetstra, R., Nijenhuis, J., Ellis, N., & van Ommen, J.R. (2009). The influence of the particle size distribution on fluidized bed hydrodynamics using high‐throughput experimentation. AIChE journal, 55(8), 2013-2023.

[Chew2015] Chew, J. W., Cahyadi, A., Hrenya, C. M., Karri, R., & Cocco, R. A. (2015). Review of entrainment correlations in gas–solid fluidization. Chemical Engineering Journal, 260, 152-171.

[Karimipour2011] Karimipour, S., & Pugsley, T. (2011). A critical evaluation of literature correlations for predicting bubble size and velocity in gas–solid fluidized beds. Powder Technology, 205(1-3), 1-14.

[Maurer 2014] Maurer, S., Schildhauer, T. J., van Ommen, J.R., Biollaz, S. M., & Wokaun, A. (2014). Scale-up of fluidized beds with vertical internals: Studying the sectoral approach by means of optical probes. Chemical Engineering Journal, 252, 131-140.

[Maurer 2015] Maurer, S., Rüdisüli, M., Teske, S.L., Schildhauer, T.J., van Ommen, J.R., Biollaz, S. M., & Wokaun, A. (2015). Transformation of local bubble rise velocity measurements to global results: Shown by a Monte Carlo simulation of a fluidized bed. International Journal of Multiphase Flow, 70, 89-95.

[Maurer 2016] Maurer, S., Gschwend, D., Wagner, E.C., Schildhauer, T.J., van Ommen, J.R., Biollaz, S.M., & Mudde, R.F. (2016). Correlating bubble size and velocity distribution in bubbling fluidized bed based on X-ray tomography. Chemical Engineering Journal, 298, 17-25.

[Rüdisüli 2012a] Rüdisüli, M., Schildhauer, T.J., Biollaz, S. M., & van Ommen, J.R. (2012). Evaluation of a sectoral scaling approach for bubbling fluidized beds with vertical internals. Chemical engineering journal, 197, 435-439.

[Rüdisüli2012b] Rüdisüli, M., Schildhauer, T.J., Biollaz, S. M., & van Ommen, J.R. (2012). Radial bubble distribution in a fluidized bed with vertical tubes. Industrial & Engineering Chemistry Research, 51(42), 13815-13824.

[Rüdisüli 2012c] Rüdisüli, M., Schildhauer, T. J., Biollaz, S. M., & van Ommen, J.R. (2012). Monte Carlo simulation of the bubble size distribution in a fluidized bed with intrusive probes. International Journal of Multiphase Flow, 44, 1-14.

[Schildhauer2016] Schildhauer T.J. (2016). Methanation for SNG production – Chemical reaction engineering aspects. In Synthetic Natural Gas from Coal, Dry Biomass, and Power-to-Gas Applications, T.J. Schildhauer, S.M.A. Biollaz (Eds.), Wiley & Sons, New York, 77-159

[Schillinger 2015] Schillinger, F., Maurer, S., Wagner, E.C., Biollaz, S.M., van Ommen, J.R., Mudde, R.F., Schildhauer, T.J. (2015). Influence of the internals Orientation on the Hydrodynamics in Gas-Solid Fluidised Beds, Proceedings of the 12th International Conference on Fluidized Bed Technology.

[Schillinger 2017] Schillinger, F., Maurer, S., Wagner, E.C., van Ommen, J.R., Mudde, R.F., Schildhauer, T.J. (2017). Influence of vertical heat exchanger tubes, their arrangement and the column diameter on the hydrodynamics in a gas–solid bubbling fluidized bed. International Journal of Multiphase Flow 97, 46-59 (2017)

[Schillinger 2018] Schillinger, F., Schildhauer, T.J., Maurer, S., Wagner, E., Mudde, R.F., & van Ommen, J.R. (2018). Generation and evaluation of an artificial optical signal based on X-ray measurements for bubble characterization in fluidized beds with vertical internals. International Journal of Multiphase Flow.

[Seemann 2010] Seemann, M. C., Schildhauer, T.J., & Biollaz, S.M.(2010). Fluidized bed methanation of wood-derived producer gas for the production of synthetic natural gas. Industrial & engineering chemistry research, 49(15), 7034-7038.

[vanOmmen2007] van Ommen, J.R., Nijenhuis, J., van den Bleek, C. M., & Coppens, M. O. (2007). Four ways to introduce structure in fluidized bed reactors. Industrial & Engineering Chemistry Research, 46(12), 4236-4244.