Modeling of a pilot-scale trickle bed reactor for the catalytic oxidation of phenol

ARC Centre for Functional Nanomaterials, University of Queensland, Brisbane QLD 4072, Australia
Separation and Purification Technology (Impact Factor: 3.07). 06/2009; DOI: 10.1016/j.seppur.2009.03.021

ABSTRACT A mathematical model was developed to simulate the catalytic wet air oxidation (CWAO) of aqueous phenol in a trickle bed reactor (TBR). Both ‘axial dispersion’ and ‘plug flow’ models were proposed. ‘Steady-state’ mass transfers across different phases inside the reactor have all been considered in parallel with oxidation reactions catalyzed by heterogeneous copper catalyst supported on activated carbon. The changes in the concentrations of oxygen and phenol in various phases were thus depicted as a function of bed length. In order to validate the accuracy of the established TBR model, a series of experiments on phenol oxidation were performed on a pilot-scale TBR containing 5.6 l of catalysts. The model was found able to give satisfactory predictions for nearly half of all the runs. The discrepancies between the experimental and modeling results were investigated for the less promising runs. It was also noticed that similar simulation results could be attained from ‘axial dispersion’ model against ‘plug flow’ model. Following the discussion on the changes of phenol and oxygen concentrations in the various phases, it is finally concluded that the performance of the TBR of this study depends largely on gas-to-liquid mass transfer process. Further suggestions with regards to reactor optimization are also proposed on the basis of experimental outcome.

  • [Show abstract] [Hide abstract]
    ABSTRACT: Background Catalytic wet air oxidation (CWAO) of phenol in a trickle-bed reactor has been investigated (operating parameters: P = 1.85 MPa, T = 393, 413, 433K, L = 0.17–1.77 kg/m2s and G=0.2–0.62 kg/m2s ). Activated carbon was used as a catalyst. The experimental results have been simulated by Eulerian multifluid model (CFD) and for the sake of comparison by the plug flow model.ResultsThe experiments proved that active carbon can be applied as a catalyst in the phenol oxidation processes revealing a catalytic activity comparable to that of commonly used metal oxides. The highest conversion of phenol obtained in long-lasting experimental runs carried out at stable catalytic activity (PO2=0.31 MPa, T = 413 K) was about 60%. Computed results show that the CFD model approximates better the experimental data (mean error 11.6%; standard deviation 5.2%) than the plug flow model (mean error 16.9%; standard deviation 12.2%).ConclusionCWAO of phenol over active carbon as a catalyst cannot be applied as an independent purification method. To remove phenol completely CWAO must be combined with other depolluting technology. The CFD model describes better the hydrodynamic phenomena in a bed of variable porosity than the simple plug flow model.
    Journal of Chemical Technology & Biotechnology 12/2014; DOI:10.1002/jctb.4610 · 2.49 Impact Factor
  • [Show abstract] [Hide abstract]
    ABSTRACT: The hydrodynamics of countercurrent gas and liquid flows through a trickle bed reactor with unstructured (8 mm Raschig rings) and structured (Bale) packing has been investigated using the residence time distribution (RTD) approach. The operation was carried out with gas and liquid Reynolds number over the range 0 < ReG < 150 and 0 < ReL < 1500, respectively. Gas phase Peclet number, PeG, was generally higher in the randomly packed Raschig rings than in the structured Bale packing but increased with ReG. However, gas phase mixing decreased with increasing liquid flow rate. Specifically, the dependency on ReG and ReL was given by, PeG = AGReGbG exp(−dReL). The gas phase hold-up, H0G also increased with gas flow rate but decreased with the liquid flow rate. Owing to the statistical nature of the gas flow through beds, a Snedecor F-distribution was used to describe the gas phase RTD data and hence, the dynamics of the macroscopic flow pattern within the bed was determined from the intensity function, I(t). This indicated initial gas recirculation in stagnant zones within the bed with the flow pattern subsequently evolving into predominantly a channelling and bypassing type. Interestingly, the liquid phase mixing behavior appeared to be independent of ReG and thus, PeL,Raschig = AL ReLbL for the unstructured bed, but flow through the more open Bale packing revealed two types of mixing regimes, with a crossover from the region of decreasing PeL to increasing PeL with an increased liquid flow rate at ReL of about 750 and accordingly, PeL,Bale = A0 + A1ReLA2ReL2. Liquid hold-up, H0L, could be decoupled into a static component, HSL, and dynamic contribution, HDL, with the latter being proportional to the liquid flow rate. The Bale packing has a higher static hold-up (0.225) than the randomly packed Raschig rings (0.100) due to its more open structure. On the whole, it seemed that liquid phase back-mixing was generally lower in the unstructured packed bed than in the structured Bale packing. The associated higher static hold-up and reduced sensitivity with increasing liquid flow rate makes the latter the preferred choice for catalytic distillation for operations bounded by the loading and flooding envelopes.
    Industrial & Engineering Chemistry Research 08/2011; 51(4):1647–1662. DOI:10.1021/ie200708y · 2.24 Impact Factor
  • [Show abstract] [Hide abstract]
    ABSTRACT: Regeneration of granular activated carbon loaded with cyclohexane by supercritical carbon dioxide was investigated at varying temperatures and pressures. Cyclohexane concentration-time profiles were measured by means of a supercritical extraction apparatus coupled with an online GC analyzer. The maximum yield of regeneration was 68 % at 55 °C and 140 bar. A mathematical model was proposed which agreed well with the experimental data. Both the Toth and the Langmuir-Freundlich isotherms were used as adsorption isotherms under equilibrium condition. The mathematical model for activated carbon regeneration was able to adjust the experimental data with relative absolute average deviations of about 4 % and 3.25 %, respectively, for the Toth and the Langmuir-Freundlich isotherms.
    Chemical Engineering & Technology 02/2013; 36(2). DOI:10.1002/ceat.201200324 · 2.18 Impact Factor