(97b) Advanced Measurement Techniques for Local Validation of Multiphase CFD Models Predictions | AIChE

(97b) Advanced Measurement Techniques for Local Validation of Multiphase CFD Models Predictions

Authors 

Uribe, S. - Presenter, Missouri University of Science and Technology
Qi, B. - Presenter, Missouri University of Science and Technology
Farid, O. - Presenter, Missouri University of Science and Technology
Al-Dahhan, M. - Presenter, Missouri University of Science & Technology-Rolla
Most of the chemical engineering systems of interest such as fluidized beds, bubble columns, slurry bubble columns, packed bed reactors and trickle bed reactors, have a prevailing multiphase, multiscale and multiphysics nature. The complexity of the multiphase and multiscale interactions of these systems represents a major challenge in the proper understanding and prediction of the flow structures and hydrodynamic behaviour. An accurate understanding and prediction of the hydrodynamic parameters on these systems is of paramount importance in order to develop scale-up methodologies and to enhance the reactors throughput. For most of these studied multiphase systems, it can be seen that with different approaches and though the application of different measurement techniques, vast experimental studies have been conducted through the last decades on these systems, allowing the characterization of the macroscopic hydrodynamics phenomena. However, in most of the reported experimental works, important underlying assumptions and limitations can be identified, and hence, uncertainties such as the presence of systematic differences by the applied measurement technique, have been also reported. Furthermore, the experimental approaches usually fail to succeed in the analysis of the microscale phenomena and their impact in the overall reactor behaviour.

In response to these limitations in the experimental studies of multiphase systems, numerical simulation though the implementation of Computational Fluid Dynamics (CFD) techniques has arisen as a promising alternative. The application of these mathematical modelling tools allows to access high resolution pointwise fields inside the systems and their changes at local times. However, most of the current mathematical descriptions of multiphase reported on literature rely on the inclusion of a series of sub-models as closures to account for the multiphase and multiscale interactions. Therefore, the predictive quality of these models is constrained by the applicability of these sub-models, which are usually a set of empirical or semi-empirical correlations or expressions, and further uncertainties arise if these models are desired to be used for extrapolation studies.

Hence, it can be noted that despite that mathematical modelling of multiphase systems can overcome the limitations of the experimental techniques, there is a fundamental need to validate the models’ predictions, and assess their predictive quality, which can be achieved by linking models with reliable experiments. In this sense, it should also be pointed out that scale-up, optimization extrapolation studies, studies on the implementation of new processes and process design, as well as the study the local behaviours inside the systems can only be performed with models with validated predictive quality.

Over the years, in our laboratory, the Multiphase Flows and Reactors Engineering and Applications Laboratory (mFReal), several invasive and non-invasive advanced measurement techniques have been developed. These in-house developed techniques allow to obtain high resolution time-resolved local fields of multiphase systems, and have been successfully applied on the studies of Trickle Bed Reactors, slurry bubble columns, gas-solid fluidized beds, moving bed reactors, among other applications. The non-invasive techniques include radioactive particle tracking (RPT), dual source gamma ray tomography (DSCT), gamma ray densitometry (GRD) for 3D flow field, velocity and turbulent parameters, phases distribution and flow pattern identification; while the invasive techniques include 4 and 2-point optical fiber probe for local bubble dynamics, liquid and gas velocities, holdup and their timewise variations. These highly sophisticated techniques have allowed to obtain reliable and useful benchmarking data for the validation of the local predictions of mathematical models, allowing to assess the models’ predictability and limitations, and enabling extrapolation studies.

In this presentation, the joined mathematical modelling and experimental studies using advanced measurement techniques of two example cases are presented.

  • An Euler-two-phase (E2P) mathematical model for the prediction of the local hydrodynamic behaviour of a gas-solid fluidized bed. Simulations are conducted for a fluidized bed of 0.14 m internal diameter packed with Gerdart B glass beads particles, with an average diameter of 365 mm. The validation of the model is conducted by comparison of local radial profiles of solids holdup, solids velocity, and pressure drops, measured on a fluidized bed without internals, packed with the same solid particles, as reported on previous contributions [1,2]. The experimental measurements were conducted by using advanced measurement techniques, 2-tip optical fiber probes and differential pressure transducer.Results show that the implemented model possesses a high predictive quality, predicting pressure drops with an Average Absolute Relative Error (AARE) between 8.6% and 11.3%; solids holdup with a Root Mean Squared Deviation (RMSD) under 10%; and solids velocity with a RMSD under 30%, and Average Absolute Deviations between 0.003 m/s and 0.36 m/s.
  • Implementation of a mathematical model for a Trickle Bed Reactor, where the solid-fluid interphase description was explicitly incorporated to account for the textural characteristics of the bed, and the momentum balances for the two fluid phases were solved in the void space between the packing. Experiments were conducted using 2-tip optical fiber probes and gamma-ray computed tomography to obtain the local liquid velocity and liquid saturation profiles at different location in the column height. The results show that the CFD model can properly predict the local variation of the liquid velocity at different flow rate conditions, with an average absolute error below 18.6 %. The CFD model properly predicted the liquid maldistribution observed in the experimental measurements. Furthermore, the CFD model results allowed to study other local phenomena, such as bypass channelling and backmixing; and also allowed to determine the variations in the interaction forces between the phases.

References:

[1] H. Taofeeq, M.H. Al-Dahhan, Effect of vertical internals on the pressure drop in a gas-solid fluidized bed, Can. J. Chem. Eng. 96 (2018) 2185–2205. doi:10.1002/cjce.23299.

[2] H. Taofeeq, M. Al-Dahhan, The impact of vertical internals array on the key hydrodynamic parameters in a gas-solid fluidized bed using an advance optical fiber probe, Adv. Powder Technol. 29 (2018) 2548–2567. doi:10.1016/j.apt.2018.07.008.