Simulation of fluidized bed reactor (FBR) was accomplished for treating wastewater using Fenton reaction, which is an advanced oxidation process (AOP). The simulation was performed to determine characteristics of FBR performance, concentration profile of the contaminants, and various prominent hydrodynamic properties (e.g., Reynolds number, velocity, and pressure) in the reactor. Simulation was implemented for 2.8 L working volume using hydrodynamic correlations, continuous equation, and simplified kinetic information for phenols degradation as a model. The simulation shows that, by using Fe3+ and Fe2+ mixtures as catalyst, TOC degradation up to 45% was achieved for contaminant range of 40–90 mg/L within 60 min. The concentration profiles and hydrodynamic characteristics were also generated. A subsequent scale-up study was also conducted using similitude method. The analysis shows that up to 10 L working volume, the models developed are applicable. The study proves that, using appropriate modeling and simulation, data can be predicted for designing and operating FBR for wastewater treatment.
Advanced oxidation processes (AOPs) have emerged to be one of the alternatives for treating effluents containing very toxic organic compounds [
Operation of FBR has confirmed many advantages that include high degradation efficiency, lesser reaction time and better catalyst recirculation [
Knowledge gap in modeling, scale-up strategies, and performance prediction is among the important factors hindering the expansion of commercial water treatments by FBR-AOP technology. Full-size commercial reactors are generally complicated and costly; so before coming into a final design of any industrial reactor, some validation of the data should be done. Mimicking hydrodynamics of lab-scale data can be a promising solution to this problem. But multiphase systems are complex in nature and scale-up of multiphase processes is highly unpredictable. Proper scale-up of any reactor requires hydrodynamic similarity in different scales. In addition validation of hydrodynamic properties at different scale is time consuming and costly by mean of experimental procedures. Under these circumstances, a theoretical method would be more appropriate to tackle the challenging problems for scale-up.
Therefore, a rule for designing a reactor process regardless of the involvement of any treatment or production can be proposed like this; firstly, to get well acquainted with the hydrodynamics, kinetics, heat transfer, and mass transfer involved in the system; secondly, to numerically solve the governing equations with help of simulation tools; and thirdly, to test scale-up methods available in the literature to come up with an approximate definition of the process. Phenol is used as an indicating contaminant for this work. Phenols and phenolic compounds are the most abundant pollutants in industrial wastewater due to their wide consumption in oil refineries, pulp and paper mills, resins, and steel coke manufacturing and pharmaceutical industries [
Now, among other available simulation tools, computational fluid dynamic (CFD) simulations have been used in previous studies on AOPs for modeling chemical species transport. It is a tool that can simultaneously solve fluid dynamic equations in the course of space and time. There are some studies on use of CFD for catalytic reactors in water treatment [
Earlier results by some researchers have confirmed the significance of combining reactor hydrodynamics to predict the degradation process [
It has been revealed by some authors that batch recirculation systems are considerably preferred in degrading highly contaminated wastewater in terms of TOC reduction [
Schematic diagram of the proposed FBR-AOP system.
The voidage was found to be 0.5989 for fluidized glass particles in water. The viscosity of the phenolic water was assumed to be the same as water. The velocity of fluidization for liquid-solid fluidization system was 0.03 m/sec, calculated using the formula given by Wen and Yu [
In the above equation,
The tube region packed with heterogeneous catalyst and glass beads in the amount of 32 g/L can occupy 70% of the reactor volume. Uniform distribution of catalyst was assumed in CFD simulation. 40–90 mg/L of phenolic water will be supplied into the system through one of the inlets, while hydrogen peroxide and water were pumped into the system through another inlet. The feed solution (water + phenol) was continuously recirculated by a pump. The fluidized bed reactor model is presented in Figure
Reactor description and applied conditions for CFD simulation.
Parameter | Specification |
---|---|
Operating temperature | 298 K |
Operating time | 3600 sec |
Solution density, |
1000 kg/m3 |
Solution viscosity, |
0.008904 poise |
|
4.46 mol/m3 and 2.55 mol/m3 |
|
27 mol/m3–16 mol/m3 |
|
1 × 10−6 m2/sec |
|
1 × 10−6 m2/sec |
|
1.54 and 0.88 gm/L |
|
14 and 8 mg/L |
Amount of glass beads | 30 gm/L |
Reactor-dia. | 7 cm |
Reactor-height | 70 cm |
Reactor-volume | 2 Liter |
Operating flow rate | 0.03 m/sec |
Bed voidage, |
0.586 (unit less) |
Permeability, |
5.9 m2 |
Reactions and rate constants for complete mineralization of phenol by Fe3+/Fe2+/H2O2.
No. | Reaction | Rate constant | Reference |
---|---|---|---|
R1 |
|
1 × 10−2 M−1s−1 | [ |
R2 |
|
3.3 × 105 M−1s−1 | [ |
R3 |
|
6.3 × 101 M−1s−1 | [ |
R4 |
|
3.3 × 107 M−1s−1 | [ |
R5 |
|
8.3 × 105 s−1 | [ |
R6 |
|
4.2 × 109 s−1 | [ |
R7 |
|
3.3 × 109 M−1s−1 | [ |
R8 |
|
7.0 × 103 M−1s−1 | [ |
R9 |
|
7.0 × 103 M−1s−1 | [ |
R10 |
|
2.0 × 1010 M−1s−1 | [ |
R11 |
|
1.1 × 1010 M−1s−1 | [ |
R12 |
|
7.0 × 103 M−1s−1 | [ |
R13 |
|
1.0 × 101 M−1s−1 | [ |
R14 | Fumaric acid + |
6.0 × 109 M−1s−1 | [ |
R15 | Oxalic acid + |
1.4 × 106 M−1s−1 | [ |
R16 | Phenol + |
0.3805 M−1s−1 | [ |
R17 | Phenol + |
1.933 × 10−2 M−1s−1 | [ |
R18 | Phenol + |
1.5 × 10−2 M−1s−1 | [ |
R19 | H2O2 |
0.0012 s−1 | From this study |
R20 | TOC1 + |
6.35 × 10−5 M−1s−1 | From this study |
R21 | TOC2 + |
9.6 × 10−7 M−1s−1 | From this study |
DHCD: di-hydroxy-cyclohexa-di-enyl radical; THB: tri-hydroxy-benzene; THCD: di-hydroxy-cyclohexa-di-enyl radical.
CFD simulation described the TOC concentration profile inside the reactor module and predicted the TOC removal performance at different operating conditions for this process. The effects of pH change on TOC reduction were negligible [
Generally, application of fluidized bed in wastewater treatment technologies is complicated. In addition, there is a lack of knowledge in this area in terms of hydrodynamics and kinetics. In this study, we tried to define the hydrodynamics and kinetics in order to depict some important parameters, such as component concentration within the reactor length with the help of CFD simulation. Though several investigations have been conducted in AOPs modeling, there are very limited studies on CFD modeling for Fenton process [
The catalyst bed of SiO2 and FeOOH was subjected to fluidize in the reactor. Only half of the reactor system was into consideration for simulation purposes as the reactor was symmetrical. The designed reactor consists of a cylindrical structure with two inflow tubes and one outlet (see Figure
This study concentrates on phenol degradation in the proposed system. Reaction kinetics of phenol degradation has been investigated in several previous studies [
The mechanism is reduced from reaction R1 to R18 based on widely accepted literature. R1 to R3 represent the formation of OH radical in reaction with Fe3+ from the catalyst and R4 to R6 represent scavenging effect of hydrogen peroxide. R7 to R12 represent production of aromatic compounds. R13 to R18 represent production of acids in degradation. As a whole, R7 to R18 represent production of intermediates. Complete mineralization regardless of production of intermediates is presented by R19 to R21. Reactions participation can be classified in three basic steps. First, the initiation and termination of radicals, hydrogen abstraction, and depropagation (decay/scission) [
Furthermore, the following reaction rate equations can be derived from (
In the proposed reactor liquid (phenol water) and solid interaction (catalyst and glass beads) will result into effective degradation of phenol. Numerical simulation of liquid flow and transport in a fluidized bed reactor is challenging as the liquid flows continuously, but the catalyst bed creates a porous zone for the liquid. On this note, this liquid movement through the reactor can be described by two steps. Firstly, free-flow of the liquid through the system and, secondly, fluid flowing through fluidized-state porous system. The momentum balance for the first step can be described with incompressible Navier-Stokes equation. The stationary Navier-stokes equations describing the fluid flow in the liberal flow regions is as follows:
In the second step, the flow through the porous fluidized catalyst bed is described by a combination of the continuity equation and momentum balance equation, which is Brinkman equation [
Brinkman equation in
The dependent variables are the concentrations of the reactants and products. Inertial term for Stokes flow and porous media are taken into account in this case.
The equation can be presented as follows after simplification:
Continuity equation was used for each compound for describing mass balance in this fluidized bed reactor. Simultaneous solving of the equation is the key step to deduce the concentration profile of TOC inside the fluidized bed reactor. It was assumed in the system that the modeled species were in very low concentrations compared to solvent liquid. Fickian approach was used for the diffusion term in mass transport. The model of 19 species and 16 reactions was done based on convection and diffusion equation. The equation is as follows:
This mass balance equation occurs for both diffusion and convection in a conservative manner. It means that the terms from the conventional part
The expression term presented in equation
Phenol is degraded by reaction with hydroxyl radical. Unsteady reactions and states are considered in the system for better understanding. Isothermal condition is assumed in the reactor system, as Fenton oxidation itself is not heat consuming and there is no significant difference found between the reactor’s inlet and outlet temperature. Furthermore, isothermal assumption reduces computational demand in the simulation. The fluid flow is incompressible, as the flow is defined as laminar and the fluid flow is Newtonian. The substantial properties of the mixture (e.g., viscosity, density, etc.) are assumed to be independent of the mass fraction of the components. The porosity and void age of the fluidized bed reactor are considered to be in constant state in the simulation system. It is assumed that the reaction only takes place in the fluidized region so the reaction rate term is zero in free-flow region. The catalyst bed of this fluidized bed reactor is assumed to be fully developed and the reacting zone is from 10 cm to 65 cm giving a height of 55 cm of reacting zone (see Figure Time-dependent solution is achieved considering an appropriate time step. Due to isothermal condition, energy balance is not employed. The boundary condition for the inlet is the known value of the velocity, no slip condition for the walls of the reactor. It is assumed that the fluid is Newtonian, incompressible, isothermal, and nonreactive with constant physical properties and under laminar steady state flow. The hydrodynamic and species transport gives mass conservation equation: A constant velocity profile is assumed at the inlet boundaries; The boundary condition for the Navier-Stokes equations at the outlet reads At the inlet, At the inlet, gradients of At outlet, pressure is set to standard atmospheric pressure at 101,325 Pa (see Figure No flux condition is applied for exterior wall which is represented by the following equation- The concentration of the inlet is fixed: At the inlet,
Boundary conditions for the catalytic fluidized bed reactor length are given as follows.
(a) Schematic diagram of the reactor (dimensions are in centimeters), (b) pressure change across the reactor, (c) mesh quality size, and (d) mesh element size.
A set of 3D models using finite element method (FEM) were considered for time dependent simulation of the mentioned system. The transport equations were described by Navier-Stoke for fluid flow, Brinkman equations for porous media, and Stephan-Maxwell equations for conversion rate of reaction and convection diffusion mechanisms. Besides, the parallel sparse direct linear solver (PARDISO) algorithm was applied to combine and solve the equations. This algorithm is a direct sparse solver which supports parallel processing. Equations (
Scaling up a reactor system is possible when there are available data of velocity or pressure drop or Reynolds number, and so forth, in different positions of the system. From CFD data of velocity, information has been gathered with change in reactor height of the proposed system. For successful scale-up of a system, a number of scale independent factors are required. However, when reactions are involved in the system it becomes quite difficult to maintain similar scale independent factors as mass transfer is highly scale dependent. The basic purpose is to obtain a set of
Again, to study the scale-up of this FBR similitude method developed by Glicksman and his coworkers was used. Although this method was developed for the scale-up of
The CFD simulations of the fluidized bed reactor were established with COMSOL multiphysics. The free flow velocity profile was achieved by simultaneous solution of continuity and momentum equations along with their respective boundary conditions. Hydrodynamic results were obtained and reaction kinetics was implemented in a developed hydrodynamic condition.
The 3D-model of the fluidized bed reactor was first solved for a time dependent laminar flow in absence of any reaction. Figure
CFD simulation of velocity at (a) time = 150 seconds, (b) time = 350 seconds, (c) time = 550 seconds, (d) time = 750 seconds, (e) time = 1000 seconds, and (f) time = 2000 seconds.
CFD simulation of (a) Reynolds number stationary-state volume plot and (b) slice plot for velocity in stationary state.
No external velocity restrictions or internal force field were positioned on the outlet, inlet, and the gradients of all variables, except for pressure, which was set to zero in the flow direction at the outlet. The inlet velocity ranged from 0.03 to 0.05 ms−1, which corresponded to a flow rate of 3–5 L/min. All inlet velocities tangential to the inlet plane were set to zero. It can be seen from the 3D velocity progression along the reactor axis that the flow was moderately homogeneous in the catalytic area governed by Brinkman equation (see Figure
A 1D line was cut from the 3D reactor volume at the middle point in the reactor system for better understanding of normal axial velocity distribution (see Figure
A one-dimensional view for velocity profile through the reactor.
A one-dimensional view for Reynolds number through the reactor.
Our aim is to degrade phenol in the fluidized state of catalyst and solution inside the reactor. For this purpose, the hydrodynamics of the reactor was simulated in absence of any reaction, and then the reactions were introduced to the system. There are 21 reactions involved in phenol degradation (see Table
A one-dimensional view for concentration profile through the reactor.
A one-dimensional plot for CFD simulation for TOC along with axial axis in mol/m3 (integrated value).
Concentration profiles of phenol (TOC1), intermediates (TOC2), and hydroxyl radicals (OH) were obtained by solving governing mass transfer equations (see (
3D plot for CFD simulation for TOC2 (intermediate formation) concentration profile in mol/m3 (a) for initial TOC of 4.46 mol/m3 and (b) for initial TOC of 2.55 mol/m3.
In these reactions, the increase in intermediate concentration was followed by a decrease in phenol concentration. This was followed by an increase in hydroxyl radical production. The life cycle of hydrogen peroxide was around 3000 sec. The CFD results gave two second order reaction rates for intermediate formation and end product conversion. It was apparent from our findings that the conversion of the reaction increased with reaction time until the equilibrium point, which reached the maximum of 44 to 48%. But in the simulated reactor, the conversion was 48% and 46%, respectively; which had 9% and 4.16% deviations (see Table
CFD result and batch study result.
No. | Experimental condition | TOC conversion | |
---|---|---|---|
Experimental study | Prediction by CFD for FBR-AOP | ||
01 | Phenol: 70 mg/L or |
44% | 48% |
|
|||
02 | Phenol: 50 mg/L or |
48% | 46% |
Physical properties and operating conditions simulated reactor and potential reactors for scale-up.
Name |
|
Column Diameter, |
Column Height, |
Liquid density, |
Liquid viscosity, |
Solid density, |
Particle diameter, |
Minimum fluidization velocity, |
Initial liquid velocity, |
Volume flow rate, |
---|---|---|---|---|---|---|---|---|---|---|
Simulated reactor | 1 | 0.07 | 0.7 | 1000 | 0.0008 | 1600 | 2 | 0.02 | 0.05 | 0.61 |
RCTR1 | 1.5 | 0.105 | 1.05 | 1000 | 0.0008 | 1600 | 2 | 0.025 | 0.061 | 0.747 |
RCTR2 | 2 | 0.21 | 2.1 | 1000 | 0.0008 | 1600 | 2 | 0.035 | 0.087 | 1.057 |
RCTR3 | 2.5 | 0.5247 | 5.247 | 1000 | 0.0008 | 1600 | 2 | 0.055 | 0.137 | 1.671 |
RCTR4 | 3 | 1.575 | 15.75 | 1000 | 0.0008 | 1600 | 2 | 0.095 | 0.237 | 2.893 |
RCTR5 | 3.5 | 5.51 | 55.1 | 1000 | 0.0008 | 1600 | 2 | 0.18 | 0.444 | 5.413 |
Five reactors were examined for scale-up from the simulated reactor. These five reactors were
Dimensionless parameters corresponding to Table
Tag name |
|
|
|
|
|
|
|
---|---|---|---|---|---|---|---|
Simulated reactor | 0.003641 | 0.625 | 2.5 | 0.285714 | 20000 | 125 | 4375 |
RCTR1 | 0.003641 | 0.625 | 2.5 | 0.190476 | 20000 | 153.0931 | 8037.388 |
RCTR2 | 0.003641 | 0.625 | 2.5 | 0.095238 | 20000 | 216.5064 | 22733.17 |
RCTR3 | 0.003643 | 0.625 | 2.5 | 0.038117 | 20000 | 342.3266 | 89809.38 |
RCTR4 | 0.003641 | 0.625 | 2.5 | 0.012698 | 20000 | 592.9271 | 466930.1 |
RCTR5 | 0.003642 | 0.625 | 2.5 | 0.00363 | 20000 | 1109.265 | 3056025 |
Screening of the reactors based on their hydrodynamic similarity with simulated reactor.
In the above, Figure
Comparison of conversion factor at different reactor lengths with changing flow rate.
The scaling up process of the simulated reactor revealed that reactor diameter has no significant effect on the reactor performance for phenol degradation. Actually, no change in hydrodynamics was observed with change in reactor diameter, given with constant reactor height. The hydrodynamic behaviour did not change with increasing or decreasing the reactor diameter for the observed numbers of m. Conversely, reactor performance was increased with lowering of initial phenol or TOC concentration. It can be explained by the fact that the reaction is of second order and the rate constant is distinctively slower; thus reactor performance could be enhanced only by decreasing the pollutant input. And also, generally systems dominated by reactions do not need strict maintenance of a bed diameter.
For having almost similar hydrodynamics as the simulated reactor (
Reactor performance of similar hydrodynamic reactors (
Treating recalcitrant aromatic compounds by various conventional treatment methods is challenging. Phenol water is hazardous to our environment; thus phenol has been chosen as a model contaminate for this work. A fluidized bed reactor process is proposed here for Fenton degradation of phenol water (40–90 mg/L). A hydrodynamic study was performed to describe phenol degradation in a 2.8 L volume catalytic fluidized bed reactor. The velocity change data in this simulation was used to produce scaled up values (using ( Brinkman equation and classical Navier stokes equation was used to describe flow through reactor geometry. Convection-diffusion equations have been used to describe the mass transfer of the reactant species. The attained graphical illustrations derived from the simulation gave a clear view on the changing velocity and Reynolds number inside the FBR. The Reynolds number through the reactor scale was in a range of Reaction mechanism of phenol has been simplified taking into account the surrogate parameter TOC. The simplified reaction that represents phenol degradation took place in the defined fluidized area. The kinetic rate constants for reactions ( From the scale-up study it is shown that up to 3 times the size of the simulated reactor similarity on hydrodynamics and performance can be attained. This established procedure can be applied to other pollutant degradation for reactor systems similar to this work. In this way, prediction of final result can be done which can be a prominent tool in design and scale-up of reactor systems involving pollutant degradation.
Thus, the monitoring of the fluidization process can be done. Consequently, it can be concluded that application of CFD simulations for the fluidized bed catalytic reactor could address a better design or extrapolation of such water treatment devices and allow a better understanding of the physicochemical phenomena involved in water treatment processes. Therefore, the stated procedure is suitable for certain design of such fluidized bed reactors for predicting performance and scale-up.
The authors declare that there is no conflict of interests regarding the publication of this paper.
The authors thank financial support from University of Malaya High Impact Research Grant (HIR-MOHE-D000037-16001) from the Ministry of Higher Education Malaysia and University of Malaya.