CFD Study of an Evaporative Trickle Bed Reactor: Mal-Distribution and Thermal Runaway Induced by Feed Disturbances
Abstract and Figures
A numerical study was carried out to investigate steady-state and transient phase distribution, evaporation, and thermal runaway in a large-scale high-pressure trickle bed reactor. A cooling recycle stream, containing reaction products and a fresh feed, was included via a closed loop calculation. It was found that, as expected, phase distribution in the catalyst bed had a substantial impact on production rate; a faulty feed distribution system can cost approximately 20% in overall steady-state product conversion. In the event that the cooling recycle stream is lost, the external reactor shell temperature can exceed its design intent. It was found that reducing the quantity of fresh reactant feed in this situation can dramatically reduce the potential for vessel damage. Thermal inertia of the catalyst particles proved to be a significant contribution to the transient energy balance. Model results are supported with a posteriori thermal excursion plant data.
Figures - uploaded by Wayne Strasser
Author content
All figure content in this area was uploaded by Wayne Strasser
Content may be subject to copyright.
... Thermal runaway of exothermic reactions generally refers to uncontrolled self-heating that is caused by cooling system failure [3] , feeding error, reactant accumulation [4] , appearance of hot spots [5,6] , mixing failure, human error, or system parametric sensitivity [7,8] , which may lead to a self-perpetuating cycle between the increase in temperature and acceleration of reaction rate. ...
Reaction thermal runaway has been extensively characterized as a major hazard for fine chemical industry. It is necessary to develop safety technologies for the control of reaction thermal runaway in emergencies and mitigating the subsequent hazards. To date, literature review on the loss prevention methods of chemical reaction thermal runaway is insufficient. In this paper, a comprehensive review is delivered to outline the progress of emergency response technologies for reaction thermal runaway in recent years, major principles and potential applications of those loss prevention methods. It is expected that this review article has certain reference value for the further understanding of thermal runaway, the design of mitigation systems, and the formulation of emergency response strategy for runaway reactions.
... Dakkoune et al. [2] investigated the causes of 169 chemical accidents in France over 40 years and found that 25% of the accidents were caused by the thermal runaway of exothermic reactions. Thermal runaway of chemical processes mainly refers to uncontrolled self-heating, which is caused by reactor cooling system failure [3], remaining reactant accumulation [4], local hot spots in reactors [5], personnel misoperation [6], or system parametric sensitivity [7]. The kettle-type reactors commonly used in large-scale industrial production are especially vulnerable as these types of reactors have an obvious amplification effect. ...
Reaction thermal runaway, caused by excessive temperatures of the reaction system, threatens the safety of operators. Latent heat storage by phase change materials (PCMs) has the advantages of high energy storage density and stable temperature during the energy storage process, which was widely applied in many fields and provides a new idea for the temperature control of thermal runaway reactions. In this study, microencapsulated phase change materials (microPCMs) with a melamine-formaldehybe (MF) resin shell was fabricated by in situ polymerization. The characterization of the micro morphology, chemical bonds, crystal structure, thermal properties, and thermal stability of microPCMs showed that the prepared microPCMs had integrated spherical morphologies and smooth surfaces, with an encapsulation ratio of approximately 70% and good thermal stability. Furthermore, taking the esterification of propionic anhydride (PA) and 2-butanol (2B) as examples, n-octadecane@MF resin microPCMs was used to control the reaction temperature under various operation conditions in semi-batch reactors. The experimental results showed that the mechanism of the n-octadecane@MF resin microPCMs on the control of reaction temperature in semi-batch reactors was the combination of both physical and chemical interactions. The applications of microPCMs for the control of reaction temperature hold great potential for use in industrial processes.
... The arrangement of the packing directly influences the flow through a packed bed and could be partially responsible for an undesired behaviour of the system (e.g. liquid maldistribution [8][9][10], especially so called "wall flow"). Besides, such information is of a great importance for validation of numerical methods designed for reconstruction of a packed bed structure, some theoretical flow models and determination of other bed properties like tortuosity [11][12][13]. ...
Detailed studies of a random packed bed structure typically require complex and expensive experimental methods. This is the main reason why such studies are rather scarce in the literature. In this
paper, a new simple method is proposed that can be used in extensive
studies of orientation distribution of particles with one axis of
symmetry, cylindrical in particular. The idea of the method is to fill
transparent particles with two immiscible substances, one with relatively low melting point above the ambient temperature (e.g. water and paraffin). The particles should be dumped in the container when the substances are in the liquid state. Then, at ambient conditions, the inclination of the interface between the solid and liquid phase represent the original orientation of a particle inside the packed bed. Sample results obtained with this method show quite good agreement with numerical simulations and other experimental works employing much more advanced techniques.
... It is thought that temperature runaway accidents are strictly related to the mechanism of chemical reactions, the activity of catalyst structure, and the sensitivity of operation parameters, etc. With the development of computer technology, CFD numerical simulation of reactor featuring considerable calculation precision and efficiency on the aspects of fluid flow, heat transmission, analog computation of chemical reactions, etc., may primarily reduce actual time and costs spent on unit experiments [8]- [10]. Anthony G. Dixon carried out a CFD numerical simulation of the heat transmission process of the bed under the influence of high Reynolds number [11]. ...
Ethylene epoxidation reactor is a kind of critical equipment in an oxidation unit. Due to the complex coupling effect of the extreme operating environment of high temperature, high pressure and potential risk factors, the reactor tends to work abnormally under the impact of parameter fluctuation, and temperature runaway incident may occur and cause severe loss. For studying the influence of parameter fluctuation on ethylene epoxidation reaction process and preventing temperature runaway accidents, physical modeling is performed firstly in this paper to acquire the chemical reaction mechanism therein and confirm the operation parameters which should be focused. The unsteady simulation of the operation process under normal situation has been performed. On such basis, a test of single-factor effect is performed to analyze the impact of parameter fluctuation on the temperature balance and operation efficiency of the reactor. Comparison analysis has been carried out between the simulation data acquired according to the established model and real production data, and it was found that the result is consistent. The analysis result indicates: higher ethene concentration, oxygen concentration, feed flux, feed temperature, drum temperature, operating pressure or smaller porosity may improve bed temperature and accelerate reaction rate; therein, ethene concentration, oxygen concentration, drum temperature, and porosity are operation parameters that need special monitoring. Based on the risk analysis of the ethylene epoxidation reactor and the simulation of the impact of parameter fluctuation, this paper studies the cause of the reactor’s temperature runaway to provide a technical basis for the forecast of reactor’s temperature runaway, early warning, and high-efficient operation of equipment. Model established based on abnormal condition parameters fluctuation, can contribute to early warning reactor temperature runaway, revealing the influence mechanism of parameter fluctuation for the reaction system, monitoring reaction process, eliminating temperature runaway in the bud, and providing theoretical and technical basis for petrochemical equipment fault diagnosis, accident prevention and control and safely long-time running.
... In addition, an adequate prediction of heat transfer rates is needed when using laboratory and bench scale TBRs to analyze the behavior of a given catalyst due to the usual requirement to operate isothermally aiming at facilitating the analysis of the experimental results [5]. It is also worth mentioning that non-stable hot spots can certainly arise both in industrial reactors [6] and laboratory reactors under fully controlled conditions [7]. ...
Heat transfer plays an important role in several applications of packed bed reactors with cocurrent downflow of liquid and gas (widely known as trickle-bed reactor - TBR).A literature survey shows that the amount of articles dealing with the prediction of heat transfer rates between a TBR and an external heating or cooling source is limited for spherical catalyst pellets and definitively scarce for other pellet shapes as cylinders and multilobes.Results from an experimental program devoted to study heat transfer between a TBR and an external jacket, employing spherical and cylindrical particles and a commercial trilobe pellet, are presented. A wide range of gas (air) and liquid (water) flow rates were covered corresponding to low and high interaction regime. A two dimensional pseudohomogeneous model was employed to represent the thermal behavior of the packed bed. Values of the effective radial thermal conductivity and the wall heat transfer coefficient were obtained by regression of radial temperature profiles for three different bed lengths. Finally, expressions to estimate both parameters for the different particle shapes were developed, thus providing a useful predictive tool, not available in the literature up to the best of our knowledge.
Esterification reactions between anhydrides and alcohols catalyzed by sulfuric acid have broad applications in the food, cosmetic, and pharmaceutical industries. However, the exothermic behavior cannot be well explained, especially in semibatch reactors with different feeding procedures. In this study, a series of esterification processes in semibatch systems were conducted with a wide range of feeding rates with two different substrates. The structures of initial catalytic substrates were characterized using in situ Fourier Transform Infrared (FTIR) spectroscopy. A total of 11 reaction pathways, including those for the initial activation of substrates and the subsequent reactions of the feeding process, were calculated by the density functional theory (DFT) method. Three possible catalytic cycles of the esterification with specific acid catalysis, protonated intermediate, and protonated intermediate and H2O, were established for the semibatch esterification reaction based on calculated results and experimental evidence. Thermal safety parameters such as reaction enthalpy (ΔHr), adiabatic temperature rise (ΔTr,ad), and maximum temperature of the synthetic reaction (MTSR) were determined based on Wilson equation used to calibrate mixing heat. The results reveal that mixing heat was 2.5 kJ/mol with same mole ratio of n-butanol (nB) and propionic anhydride (PA). The reactions started by PA are dominant, no matter whether the substrate is nB or PA. The initial reaction rate is restricted by the low concentration of catalyst and active reactant when nB is used as substrate, which lead to significant reactant accumulation with high MTSR of approximately 160 °C. Furthermore, feeding nB into PA allows a controllable increase in the ΔTr,ad realized by changing the feeding rate and limiting the accumulation of reactant.
Reaction inhibition is a promising option as an emergency measure to mitigate hazards caused by exothermic runaway reactions. The applications of phase change materials (PCMs) as inhibitors of runaway reactions have been influenced by thermal energy storage technologies. In this work, nano-PCMs powders with silica shells were fabricated using the sol-gel method at alkaline conditions. The prepared nano-PCMs had perfect core-shell microstructures and spherical morphologies, with an average encapsulated PCM particle diameter of 717.34 nm. In addition, the nano-PCMs exhibited high latent heats of fusion (129.1 J g⁻¹), which could inhibit the uncontrolled self-heating caused by runaway reactions through conversion of the heats of reaction to the latent heats of PCMs. The homogeneous esterification of propionic anhydride with n-butanol catalyzed by sulfuric acid was chosen as target reaction with risk of exothermic runaway. Inhibition experiments was conducted in a batch reactor coupled with an in situ FTIR spectrometer. The heat transfer performance of reactant flow was studied to analyze the mechanism of thermal runaway inhibition with nano-PCMs. The results revealed that thermal runaway can be halted by applying nano-PCMs and that the reaction can proceed to completion steadily. The effect of Re on heat transfer performance was enlarged with addition of nano-PCMs. The enhancement in the heat transfer coefficient for PCM slurry was proved. Furthermore, the effectiveness of the thermal runaway inhibition can be significantly influenced by the warning temperature, stirring rate, and mass of added nano-PCMs. The applications of nano-PCMs for runaway reaction inhibition holds great potential for use in industrial processes.
The paper is devoted to interfacial heat exchange between gas and liquid flowing countercurrently through a packed bed in a trickling flow regime. The lack of correlations describing the interfacial gas-liquid heat transfer coefficient makes problems when numerical models of non-isothermal flows in porous media are being developed. Thus the experimental investigation was undertaken with the use of a column of 0.1 m inner diameter, equipped with 6 mm glass Raschig rings. Air and water were used as working fluids. The loads of media ranged between 0.0177–0.1415 m3·(m2 s)−1 and 0.0007–0.0053 m3·(m2 s)−1 for gas and liquid phases, respectively. The inlet water temperature was changed within the range between 30 °C and 70 °C whereas the inlet air temperature was kept constant at the level of (21 ± 1) °C. It was found that interfacial heat transfer coefficient is strongly dependent on the gas load, noticeably dependent on the temperature difference between phases and slightly dependent on the liquid load. The results of the experiment were used to develop a new correlation describing the interfacial heat transfer in the packed bed expressed by the Nusselt number. Various group numbers were considered in order to account for the impact of gravity, surface tension, thermal diffusion and free convection on the interfacial heat transfer. After detailed regression analysis the correlation of the form Nu=ReG^1.169·GaG^-0.8399·Eo^0.7176 was finally proposed as the most fitting the experimental data.
Loss mechanisms in a scallop shrouded transonic power generation turbine blade passage at realistic engine conditions have been identified through a series of large-scale (typically 12 million finite volumes) simulations. All simulations are run with second-order discretization and viscous sublayer resolution, and they include the effects of viscous dissipation. The mesh (y+ near unity on all surfaces) is highly refined in the tip clearance region, casing recesses, and shroud region in order to fully capture complex interdependent flow physics and the associated losses. Aerodynamic losses, in order of their relative importance, are a result of the following: separation around the tip, recesses, and shroud; tip vortex creation; downstream mixing losses, localized shocks on the airfoil; and the passage vortex emanating from under the shroud. A number of helical lateral flows were established near the upper shroud surfaces as a result of lateral pressure gradients on the scalloped shroud. It was found that the tip leakage and passage losses increased approximately linearly with increasing tip clearance, both with and without the effect of the relative casing motion. For each tip clearance studied, scrubbing slightly reduced the tip leakage, but the overall production of entropy was increased by more than 50%. Also the overall passage mass flow rate, for a given inlet total pressure to exit static pressure ratio, increased almost linearly with increasing tip clearance. In addition, it was also found that there was slight positive and negative lift on the shroud, depending on the tip clearance. At the lowest tip clearance of 20 mils there was a negative lift on the shroud. In the 200-mil tip clearance case there was a positive lift on the shroud. The relative motion of the casing contributed positively to the lift at every tip clearance, affecting more at the lowest tip clearance where the casing is closest to the blade tip. Lastly, it was found that the computed entropy generation for the stationary 80-mils case using the SKE turbulence model was close to that of the 80-mils scrubbing case using the RKE turbulence model. In light of the proposed mechanisms and their relative contributions, suggested design considerations are posed.
An in-depth numerical study has been carried out to investigate a high-pressure commercial scale (2–8 m diameter, 30–40 m in height) slurry bubble column reactor. Typical superficial gas velocities are in the range of 0.5–3 m/s, and overall vapor holdups are in the range of 0.45–0.85. The study revealed that steady compartmental reaction models do not match plant data when reaction time constants are fast. Also, off-the-shelf commercial computational fluid dynamics codes do not produce useful information about a reactive column of this scale without first validating the model using data “anchors” from full-scale operational columns. Important measures include both transient and time-averaged profiles, integrals, and extrema of vapor holdup and reactants. Reactor designs based on this study show both improved productivity and product quality, allowing record production from existing plants along with substantial capital scope reduction for new plants. © 2011 American Institute of Chemical Engineers AIChE J, 2012
A 3D computational fluid dynamics investigation of particle-induced flow effects and liquid entrainment from an industrial-scale separator has been carried out using the Eulerian-Lagrangian two-way coupled multiphase approach. A differential Reynolds stress model was used to predict the gas phase turbulence field. The dispersed (liquid) phase was present at an intermediate mass loading (0.25) but low volume fraction (0.05). A discrete random walk method was used to track the paths of the liquid droplet releases. It was found that gas phase deformation and turbulence fields were significantly impacted by the presence of the liquid phase; these effects have been parametrically quantified. Substantial enhancement of both the turbulence and the anisotropy of the continuous phase by the liquid phase was demonstrated. It was also found that a large number (&1000) of independent liquid droplet release events were needed to make conclusions about liquid entrainment. Known plant run conditions and entrainment rates validated the numerical method.
A moving-deforming grid study was carried out using a commercial computational fluid dynamics (CFD) solver, FLUENT® 6.2.16. The goal was to quantify the level of mixing of a lower-viscosity additive (at a mass concentration below 10%) into a higher-viscosity process fluid for a large-scale metering gear pump configuration typical in plastics manufacturing. Second-order upwinding and bounded central differencing schemes were used to reduce numerical diffusion. A maximum solver progression rate of 0.0003 revolutions per time step was required for an accurate solution. Fluid properties, additive feed arrangement, pump scale, and pump speed were systematically studied for their effects on mixing. For each additive feed arrangement studied, the additive was fed in individual stream(s) into the pump-intake. Pump intake additive variability, in terms of coefficient of spatial variation (COV), was >300% for all cases. The model indicated that the pump discharge additive COV ranged from 45% for a single centerline additive feed stream to 5.5% for multiple additive feed streams. It was found that viscous heating and thermal/shear-thinning characteristics in the process fluid slightly improved mixing, reducing the outlet COV to 3.2% for the multiple feed-stream case. The outlet COV fell to 2.0% for a half-scale arrangement with similar physics. Lastly, it was found that if the smaller unit's speed were halved, the outlet COV was reduced to 1.5%.
Previously obtained experimental heat transfer data have been collected and are illustrated along with minor variations of the standard correlations. Analysis of data for heat transfer in randomly packed beds and compact (void fraction less than 0.65) staggered tube bundles indicates that the Nusselt number for a wide range of packing materials and tube arrangements is given by
provided NRe ≥ 50. The correlations presented in this paper are not necessarily the most accurate available; however, they have wide application, are easy to use, and are quite satisfactory for most design calculations.
Unsteady flow features of a plant-scale (>1.5 m diameter) cyclone-ejector system have been studied numerically and validated experimentally. Complexity arises from the fact that the transient pressure field within the Lapple cyclone governs the operation of the annular ejector, and vice versa. Eight geometric configurations for improving the system operation were evaluated. Simple geometric changes were shown numerically to make operational improvements while incrementally improving particle collection efficiency. It was also found that compressible, time-dependent CFD results were extremely sensitive to the pressure discretisation approach and to the differential Reynolds Stress pressure strain formulation.
The present work encompasses an assessment of multiphase fluid modelling techniques to allow the prediction of reaction parameters in trickle-bed reactors (TBR). After the development of volume-of-fluid (VOF) and an Euler–Euler models, the catalytic wet air oxidation of phenolic acids was simulated under unsteady state evaluating axial and radial profiles for total organic carbon concentration and temperature for the bulk phase.For the purpose of code validation, theoretical results were compared with experimental data in terms of major hydrodynamic parameters, pressure drop and liquid holdup. The Euler–Euler model gave better predictions in comparison with VOF model since it used empirically based interphase coupling parameters in the momentum balance equation.After the hydrodynamic validation, both multiphase models were used to investigate the dynamic performance of TBR under reaction conditions for the pollutant decontamination of phenolic wastewaters. VOF exhibited the highest TOC conversion as well as the highest temperatures. The Euler–Euler model predictions gave rise also to the existence of hotspot formation in the first half of TBR being this fact related with poor radial mixing attained by means of CFD codes.
The author previously used the principle of momentum interpolation to calculate some simple one- and two-dimensional test flow situations using nonstaggered variable arrangements. An important observation during these trial runs was that the converged result for any flow situation considered depended on the underrelaxation parameter used for velocity. Here he identifies the basic reason for the observed dependence of the results on the underrelaxation parameter and proposes how to implement the momentum interpolation in an iterative algorithm to achieve a unique solution that is independent of the underrelaxation parameter used.